{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"../img/ods_stickers.jpg\" />"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 决策树和 K 近邻分类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 介绍"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "本次实验分别运用决策树和最近邻方法在分类任务上构建模型，并通过交叉验证对模型进行调优。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 知识点"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 决策树\n",
    "- 最近邻方法\n",
    "- 交叉验证"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 机器学习介绍"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在深入本次实验之前，首先了解一下什么是机器学习， Machine Learning（T. Mitchell 著，1997 年出版）一书中给出了机器学习经典、通用的定义："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">假设用 P 来评估计算机程序在某任务类 T 上的性能，若一个程序利用经验 E 在任务 T 上获得了性能改善，则我们就说关于 T 和 P, 该程序对 E 进行了学习。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在不同的问题设定下，T、P、E 可能指完全不同的东西。机器学习中一些流行的任务 T 包括："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 分类：基于特征将实例分为某一类。\n",
    "- 回归：基于实例的其他特征预测该实例的数值型目标特征。\n",
    "- 聚类：基于实例的特征实现实例的分组，从而让组内成员比组间成员更为相似。\n",
    "- 异常检测：寻找与其他样本或组内实例有很大区别的实例。\n",
    "- 其他更多任务"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "关于经验 E，《Deep Learning》（Ian Goodfellow、Yoshua Bengio、Aaron Courville 著，2016 年出版）的「Machine Learning basics」一章提供了一份很好的综述："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">经验 E 指的是数据（没有数据我们什么也干不了）。根据训练方式，机器学习算法可以分为监督（supervised）和无监督（unsupervised）两类。无监督学习需要训练含有很多特征的数据集，然后学习出这个数据集上有用的结构性质。而监督学习的数据集除了含有很多特征外，它的每个样本都要有一个标签（label）或目标（target）。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 示例"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "分类和回归属于监督学习问题。例如，作为信贷机构，我们可能希望根据客户累积的数据预测贷款违约情况。在这里，经验 E 是已有的训练数据，即实例（客户）的集合，一组特征（例如年龄、薪水、贷款类型、以往违约记录等），一个目标变量（他们是否会违约）。由于需要预测的目标变量是「他们是否会违约」，所以这是一个二元分类问题。如果你转而预测贷款会超期多久，那么需要预测的目标变量变成了一个连续值（时间），这就成为一个回归问题了。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "最后，关于算法表现的评估度量 P。不同问题和算法的度量不同，当学习新算法时，我们将讨论这一点。就目前而言，本次实验将使用分类算法中的一个简单度量标准，即准确率（Accuracy）。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面看看分类和回归这两个监督学习问题。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 决策树 "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "决策树是分类与回归问题中常用的方法之一。其实不仅是机器学习领域，在每天的日常决策中，我们都在使用决策树。流程图实际上就是决策树的可视化表示，例如，下面是俄罗斯国立高等经济研究大学（Higher School of Economics）提供的关于「如何在学院网站上发表论文」的流程图："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img width='500px;' src='https://doc.shiyanlou.com/courses/uid214893-20190505-1557025112634'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "用机器学习的术语来说，可以把它看成一个简单的分类器，根据内容（书、小册子、论文）、新闻类型、原发表物类型（科学期刊、通讯）等来确定合适的发表类型（书、文章、书的章节、预印本、Higher School of Economics and the Media 稿件）。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "决策树常常是专家经验的概括，是一种分享特定过程知识的方式。例如，在引入可扩展机器学习算法之前，银行业的信用评分任务是由专家解决的，能否放贷是基于一些直观（或经验）的规则，这些规则就可以表示为决策树的形式，如下图所示："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"https://doc.shiyanlou.com/courses/uid214893-20190505-1557025095926\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "作为机器学习算法的决策树基本上和上图差不多，它合并一连串逻辑规则，使之成为一个树形的数据结构，这些规则的形式为「特征 a 的值小于 x，特征 b 的值小于 y … => 类别 1」。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面，我们基于「年龄」、「房产」、「收入」、「教育」特征使用决策树解决一个二元分类问题，即「是否允许贷款」。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 如何构建决策树"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "年龄、房产、收入、教育，这么多的特征首先应该关注哪个呢？"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "为了回答上述问题，先看一个简单的游戏，即「20个问题」游戏，这个游戏是这样玩的：A 心里想着一个名人，B 问 A 20 个问题，A 只能回答「是」或「否」，20 个问题之后 B 要猜出 A 心里想的那个名人是谁。首先问一个可以最大程度压缩剩余选项数目的问题会使 B 占据极大优势，例如询问「是不是安吉丽娜·朱莉？」，最多剔除一个选项，而询问「这个名人是女人吗？」将消除大约一半的选项。就是说，「性别」特征相比「安吉丽娜·朱莉」、「西班牙人」、「喜欢足球」等其他特征更能区分名人数据集。这背后的道理与熵有关，下面介绍熵的概念。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 熵"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "熵是一个在物理、信息论和其他领域中广泛应用的重要概念，可以衡量获得的信息量。对于具有 N 种可能状态的系统而言，熵的定义如下："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$S = -\\sum_{i=1}^{N}p_i \\log_2{p_i},$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "其中，$p_i$ 是系统位于第 i 个状态的概率。熵可以描述为系统的混沌程度，熵越高，系统的有序性越差，反之亦然。熵将帮助我们高效的分割数据，类似帮助我们找出在「20个问题」游戏中先问什么问题较好。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 玩具示例"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "为了解释熵是如何有利于构建决策树模型的，让我们来看一个玩具示例，在这个示例中将基于球的位置预测它的颜色。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src='https://doc.shiyanlou.com/courses/uid214893-20190505-1557025079468'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "这里有 9 个蓝球和 11 个黄球。如果随机选择一个球，这个球是蓝球的概率 $p_1=\\frac{9}{20}$，是黄球的概率 $p_2=\\frac{11}{20}$，这意味着熵 $S_0 = -\\frac{9}{20}\\log_2{\\frac{9}{20}}-\\frac{11}{20}\\log_2{\\frac{11}{20}} \\approx 1$。这个值本身可能无法告诉我们很多信息。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "将球分为「位置小于等于 12、位置大于 12」这两组，如下图所示。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src='https://doc.shiyanlou.com/courses/uid214893-20190505-1557025058742'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "那么分组后，熵的变化如何？左边一组有 13 个球， 8 蓝 5 黄。这一组的熵 $S_1 = -\\frac{5}{13}\\log_2{\\frac{5}{13}}-\\frac{8}{13}\\log_2{\\frac{8}{13}} \\approx 0.96$。右边一组有 7 个球， 1 蓝 6 黄。右边这组的熵 $S_2 = -\\frac{1}{7}\\log_2{\\frac{1}{7}}-\\frac{6}{7}\\log_2{\\frac{6}{7}} \\approx 0.6$。可见，两组的熵都下降了，且右边这组降得更多。由于熵实际上是系统混沌（或不确定）的程度，熵的下降被称为信息增益。数学上，基于变量 Q（在这个例子中是变量「x ≤ 12」）所作的分割，得到的信息增益（IG）定义为："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$IG(Q) = S_O - \\sum_{i=1}^{q}\\frac{N_i}{N}S_i,$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "其中，$q$ 是分割的组数，$N_i$ 是变量 Q 等于第 i 项时的样本数目。在玩具示例中，有 2 个组（$q = 2$），一组有 13 个元素（$N_1 = 13$），另一组有 7 个（$N_2 = 7$）。因此，信息增益为："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$ IG(x \\leq 12) = S_0 - \\frac{13}{20}S_1 - \\frac{7}{20}S_2 \\approx 0.16.$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "结果表明，根据「坐标小于或等于12」将球分为两组带来了一个更有序的系统。让我们继续分组，直到每组中的球颜色都一样。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src='https://doc.shiyanlou.com/courses/uid214893-20190505-1557025043052'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图可见，右边那组只需根据「坐标小于或等于 18」再分割一次即可。而左边那组还需要三次分割。注意，若组内所有球的颜色都一样，那么这个组的熵为 0（$\\log_2{1} = 0$）。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "通过这个例子，我们成功构建了一个基于球的位置预测球颜色的决策树。但倘若我们再向里面增加一个球，这个决策树就可能无法很好地工作，因为它完全拟合了训练集（初始的 20 球）。如果希望提升它的泛用性，那么一棵具有更少分支（「问题」）的决策树将有更好的效果。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 决策树构建算法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在之前的例子中构建的决策树是最优的：它只需提 5 个「问题」（基于变量 Q），就完全拟合了训练集。其他分割条件会使得到的树更深，即需要更多「问题」才能获得答案。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "构建决策树的流行算法（如 ID3 或 C4.5）的核心，是贪婪最大化信息增益：在每一步，算法都会选择能在分割后给出最大信息增益的变量。接着递归重复这一流程，直到熵为零（或者，为了避免过拟合，直到熵为某个较小的值）。不同的算法使用不同的推断，通过「提前停止」或「截断」以避免构建出过拟合的树。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 分类问题中其他的分割质量标准"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上面我们讨论了熵是如何衡量树的分区的，但还有其他指标来衡量分割的好坏："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 基尼不确定性（Gini uncertainty）：$G = 1 - \\sum\\limits_k (p_k)^2$\n",
    "- 错分率（Misclassification error）：$E = 1 - \\max\\limits_k p_k$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "实践中几乎从不使用错分率，而基尼不确定性和信息增益的效果差不多。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " 二元分类问题的熵和基尼不确定性为："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$ S = -p_+ \\log_2{p_+} -p_- \\log_2{p_-} = -p_+ \\log_2{p_+} -(1 - p_{+}) \\log_2{(1 - p_{+})}$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$ G = 1 - p_+^2 - p_-^2 = 1 - p_+^2 - (1 - p_+)^2 = 2p_+(1-p_+)$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "其中 $p_+$ 是对象具有标签 + 的概率。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "以 $p_+$ 为坐标，绘制上面两个函数的图像。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "sns.set()\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f537d252a58>"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(6, 4))\n",
    "xx = np.linspace(0, 1, 50)\n",
    "plt.plot(xx, [2 * x * (1 - x) for x in xx], label=\"gini\")\n",
    "plt.plot(xx, [4 * x * (1 - x) for x in xx], label=\"2*gini\")\n",
    "plt.plot(xx, [-x * np.log2(x) - (1 - x) * np.log2(1 - x) for x in xx], label=\"entropy\")\n",
    "plt.plot(xx, [1 - max(x, 1 - x) for x in xx], label=\"missclass\")\n",
    "plt.plot(xx, [2 - 2 * max(x, 1 - x) for x in xx], label=\"2*missclass\")\n",
    "plt.xlabel(\"p+\")\n",
    "plt.ylabel(\"criterion\")\n",
    "plt.title(\"Criteria of quality as a function of p+ (binary classification)\")\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图可见，熵的图像和两倍的基尼不确定性图像非常接近。因此，在实践中，这两个指标的效果基本上是一样的。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 示例"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面用一棵决策树拟合一些合成数据。这些合成数据属于两个不同的类别，这两个类别的均值不同，但都呈现正态分布。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 第一类\n",
    "np.random.seed(17)\n",
    "train_data = np.random.normal(size=(100, 2))\n",
    "train_labels = np.zeros(100)\n",
    "\n",
    "# 第二类\n",
    "train_data = np.r_[train_data, np.random.normal(size=(100, 2), loc=2)]\n",
    "train_labels = np.r_[train_labels, np.ones(100)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面绘制数据。通俗地讲，这种情况下的分类问题就是构造一个「边界」，能够较好的分开两个类别（红点和黄点）。这个「边界」若是一条直线的话可能太过简单，若是沿着每个红点画出的蛇形曲线又太过复杂（这将导致其在新数据上的表现很差）。从直觉上说，某种平滑的边界，在新数据上的效果会比较好。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f537d1db2e8>]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 8))\n",
    "plt.scatter(\n",
    "    train_data[:, 0],\n",
    "    train_data[:, 1],\n",
    "    c=train_labels,\n",
    "    s=100,\n",
    "    cmap=\"autumn\",\n",
    "    edgecolors=\"black\",\n",
    "    linewidth=1.5,\n",
    ")\n",
    "plt.plot(range(-2, 5), range(4, -3, -1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面训练一棵 sklearn 决策树，区分这两类数据点。最后可视化所得的边界。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f537141a978>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "# 编写一个辅助函数，返回之后的可视化网格\n",
    "\n",
    "\n",
    "def get_grid(data):\n",
    "    x_min, x_max = data[:, 0].min() - 1, data[:, 0].max() + 1\n",
    "    y_min, y_max = data[:, 1].min() - 1, data[:, 1].max() + 1\n",
    "    return np.meshgrid(np.arange(x_min, x_max, 0.01), np.arange(y_min, y_max, 0.01))\n",
    "\n",
    "\n",
    "#  max_depth参数限制决策树的深度\n",
    "clf_tree = DecisionTreeClassifier(criterion=\"entropy\", max_depth=3, random_state=17)\n",
    "# 训练决策树\n",
    "clf_tree.fit(train_data, train_labels)\n",
    "# 可视化\n",
    "xx, yy = get_grid(train_data)\n",
    "predicted = clf_tree.predict(np.c_[xx.ravel(), yy.ravel()]).reshape(xx.shape)\n",
    "plt.pcolormesh(xx, yy, predicted, cmap=\"autumn\")\n",
    "plt.scatter(\n",
    "    train_data[:, 0],\n",
    "    train_data[:, 1],\n",
    "    c=train_labels,\n",
    "    s=100,\n",
    "    cmap=\"autumn\",\n",
    "    edgecolors=\"black\",\n",
    "    linewidth=1.5,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "通过 pydotplus 和 export_graphviz 库我们可以方便的看到决策树本身是怎样的。使用 `StringIO()` 函数开辟一个缓存空间保存决策树，通过 `export_graphviz()` 函数以 DOT 格式导出决策树的 GraphViz 表示，然后将其写入 out_file 中。使用 `graph_from_dot_data()` 函数读入数据并通过 `Image()` 函数显示决策树。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting pydotplus\n",
      "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/60/bf/62567830b700d9f6930e9ab6831d6ba256f7b0b730acb37278b0ccdffacf/pydotplus-2.0.2.tar.gz (278kB)\n",
      "\u001b[K    100% |████████████████████████████████| 286kB 7.3MB/s eta 0:00:01\n",
      "\u001b[?25hRequirement already satisfied: pyparsing>=2.0.1 in /opt/conda/lib/python3.6/site-packages (from pydotplus) (2.2.0)\n",
      "Building wheels for collected packages: pydotplus\n",
      "  Building wheel for pydotplus (setup.py) ... \u001b[?25ldone\n",
      "\u001b[?25h  Stored in directory: /tmp/.cache/pip/wheels/35/7b/ab/66fb7b2ac1f6df87475b09dc48e707b6e0de80a6d8444e3628\n",
      "Successfully built pydotplus\n",
      "Installing collected packages: pydotplus\n",
      "Successfully installed pydotplus-2.0.2\n",
      "\u001b[33mYou are using pip version 19.0.3, however version 19.1.1 is available.\n",
      "You should consider upgrading via the 'pip install --upgrade pip' command.\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "!pip install pydotplus  # 安装必要模块"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6685b7e642e840e38371422e5935c224",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Image(value=b'\\x89PNG\\r\\n\\x1a\\n\\x00\\x00\\x00\\rIHDR\\x00\\x00\\x04\\xc8\\x00\\x00\\x01\\xf1\\x08\\x02\\x00\\x00\\x00\\x17\\xe1\\…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from io import StringIO\n",
    "\n",
    "import pydotplus\n",
    "from ipywidgets import Image\n",
    "from sklearn.tree import export_graphviz\n",
    "\n",
    "dot_data = StringIO()\n",
    "export_graphviz(clf_tree, feature_names=[\"x1\", \"x2\"], out_file=dot_data, filled=True)\n",
    "graph = pydotplus.graph_from_dot_data(dot_data.getvalue())\n",
    "Image(value=graph.create_png())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图表明，在最深的一层，树将空间「切割」为8个矩形，也就是说，树有8个叶节点。在每个矩形之中，树将根据数量较多的对象的标签做出预测。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 我们如何「读懂」这颗决策树？"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上个示例中，总共有 200 个合成数据（样本），每个分类各有 100 个合成数据。初始状态的熵是最大的，即 $S=1$。接着，通过比较 $x_2$ 与 1.211 的大小进行第一次分割，将样本分成两组（你可以在上图中找到这一部分边界）。基于这一次分割，左右两组的熵都下降了。这一过程持续进行，直到树的深度达到 3。在上图中，属于第一类的样本数量越多，该节点的橙色就越深，属于第二类的样本越多，该节点的蓝色就越深。若两类样本的数量相等，则为白色，比如根节点的两类样本数量相同，所以它是白色的"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 决策树如何应用到数值特征？"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "假设有一个数值特征「年龄」，该特征有大量的唯一值。决策树将通过查看「年龄 < 17」、「年龄 < 22.87」这样的二元属性寻找最好的分割，分割的好坏由某种信息增益标准衡量。但在构建树的每一步中，会有过多的二元属性可供选择，比如「薪水」同样能以很多方式进行分割，为了解决这一问题，我们经常使用启发式算法来限制选择的属性数量。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "看一个例子，假设有如下数据集："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Age</th>\n",
       "      <th>Loan Default</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>38</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>49</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>55</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>25</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>29</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>31</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>33</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Loan Default\n",
       "0    17             1\n",
       "1    64             0\n",
       "2    18             1\n",
       "3    20             0\n",
       "4    38             1\n",
       "5    49             0\n",
       "6    55             0\n",
       "7    25             1\n",
       "8    29             1\n",
       "9    31             0\n",
       "10   33             1"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.DataFrame(\n",
    "    {\n",
    "        \"Age\": [17, 64, 18, 20, 38, 49, 55, 25, 29, 31, 33],\n",
    "        \"Loan Default\": [1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1],\n",
    "    }\n",
    ")\n",
    "data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用 `sort_values()` 方法根据年龄进行升序排列。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Age</th>\n",
       "      <th>Loan Default</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>25</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>29</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>31</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>33</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>38</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>49</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>55</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Loan Default\n",
       "0    17             1\n",
       "2    18             1\n",
       "3    20             0\n",
       "7    25             1\n",
       "8    29             1\n",
       "9    31             0\n",
       "10   33             1\n",
       "4    38             1\n",
       "5    49             0\n",
       "6    55             0\n",
       "1    64             0"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.sort_values(\"Age\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "训练一个决策树模型，并可视化。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "eaae3c582a6548d4ae7b85d513b718d1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Image(value=b'\\x89PNG\\r\\n\\x1a\\n\\x00\\x00\\x00\\rIHDR\\x00\\x00\\x01\\xf4\\x00\\x00\\x03\\x07\\x08\\x06\\x00\\x00\\x00\\xcfO\\xa5…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "age_tree = DecisionTreeClassifier(random_state=17)\n",
    "age_tree.fit(data[\"Age\"].values.reshape(-1, 1), data[\"Loan Default\"].values)\n",
    "\n",
    "dot_data = StringIO()\n",
    "export_graphviz(age_tree, feature_names=[\"Age\"], out_file=dot_data, filled=True)\n",
    "graph = pydotplus.graph_from_dot_data(dot_data.getvalue())\n",
    "Image(value=graph.create_png())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图可见，该决策树使用以下 5 个值来评估年龄：43.5、19、22.5、30、32。如果你仔细观察，你会发现它们就是目标变量（Loan Default）出现变化（从 1「切换」到 0 或从 0「切换」到 1）时那两个年龄的平均值。比如，一个 38 岁的客户没能偿还贷款（目标变量为 1），而一个 49 岁的客户还贷了（目标变量为 0），那么树使用的评估值就是 38 和 49 的均值，即 43.5。树寻找那些目标变量发生变化的值，以此作为「切割」的阈值。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面考虑一个更复杂的例子，把「薪水」变量（以千美元每年为单位）加入数据集。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Age</th>\n",
       "      <th>Salary</th>\n",
       "      <th>Loan Default</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>17</td>\n",
       "      <td>25</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>18</td>\n",
       "      <td>22</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20</td>\n",
       "      <td>36</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>25</td>\n",
       "      <td>70</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>29</td>\n",
       "      <td>33</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>31</td>\n",
       "      <td>102</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>33</td>\n",
       "      <td>88</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>38</td>\n",
       "      <td>37</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>49</td>\n",
       "      <td>59</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>55</td>\n",
       "      <td>74</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64</td>\n",
       "      <td>80</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Salary  Loan Default\n",
       "0    17      25             1\n",
       "2    18      22             1\n",
       "3    20      36             0\n",
       "7    25      70             1\n",
       "8    29      33             1\n",
       "9    31     102             0\n",
       "10   33      88             1\n",
       "4    38      37             1\n",
       "5    49      59             0\n",
       "6    55      74             0\n",
       "1    64      80             0"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data2 = pd.DataFrame(\n",
    "    {\n",
    "        \"Age\": [17, 64, 18, 20, 38, 49, 55, 25, 29, 31, 33],\n",
    "        \"Salary\": [25, 80, 22, 36, 37, 59, 74, 70, 33, 102, 88],\n",
    "        \"Loan Default\": [1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1],\n",
    "    }\n",
    ")\n",
    "data2.sort_values(\"Age\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上表可见，如果根据年龄排序，目标变量（Loan Default）将切换（从 1 到 0 或从 0 到 1）5 次。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下表可见，如果根据薪水排序，它将切换 7 次。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Age</th>\n",
       "      <th>Salary</th>\n",
       "      <th>Loan Default</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>18</td>\n",
       "      <td>22</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>17</td>\n",
       "      <td>25</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>29</td>\n",
       "      <td>33</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20</td>\n",
       "      <td>36</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>38</td>\n",
       "      <td>37</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>49</td>\n",
       "      <td>59</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>25</td>\n",
       "      <td>70</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>55</td>\n",
       "      <td>74</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64</td>\n",
       "      <td>80</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>33</td>\n",
       "      <td>88</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>31</td>\n",
       "      <td>102</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Salary  Loan Default\n",
       "2    18      22             1\n",
       "0    17      25             1\n",
       "8    29      33             1\n",
       "3    20      36             0\n",
       "4    38      37             1\n",
       "5    49      59             0\n",
       "7    25      70             1\n",
       "6    55      74             0\n",
       "1    64      80             0\n",
       "10   33      88             1\n",
       "9    31     102             0"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data2.sort_values(\"Salary\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面看看树将如何选择特征。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,\n",
       "            max_features=None, max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, presort=False, random_state=17,\n",
       "            splitter='best')"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "age_sal_tree = DecisionTreeClassifier(random_state=17)\n",
    "age_sal_tree.fit(data2[[\"Age\", \"Salary\"]].values, data2[\"Loan Default\"].values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ce7147d0e35a4c0e92b1bc75413abba8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Image(value=b'\\x89PNG\\r\\n\\x1a\\n\\x00\\x00\\x00\\rIHDR\\x00\\x00\\x01\\xfc\\x00\\x00\\x02|\\x08\\x02\\x00\\x00\\x00\\x94h\\x84\\x0…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "dot_data = StringIO()\n",
    "export_graphviz(\n",
    "    age_sal_tree, feature_names=[\"Age\", \"Salary\"], out_file=dot_data, filled=True\n",
    ")\n",
    "graph = pydotplus.graph_from_dot_data(dot_data.getvalue())\n",
    "Image(value=graph.create_png())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图表明，树同时根据薪水和年龄进行分区，有些节点的分割阈值选择了年龄，有些选择了薪水。树为何选择这些特征？因为根据基尼不确定性质量标准，它们提供了更好的分区。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "结论：决策树处理数值特征最简单的启发式算法是升序排列它的值，然后只关注目标变量发生改变的那些值。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "此外，当数据集具有大量数值特征，且每个特征具有大量唯一值时，只选择最高的N个阈值，即，仅仅使用提供最高增益的前N个值。这一过程可以看成是构造了一棵深度为 1 的树，计算熵（或基尼不确定性），然后选择最佳阈值用于比较。比方说，如果我们根据「薪水 ≤ 34.5」分割，左子组的熵为 0（所有客户都是「不好的」），而右边的熵为 0.954（3 个「不好的」，5 个「好的」，你可以自行确认这一点，这将作为作业的一部分），信息增益大概是 0.3。如果我们根据「薪水 ≤ 95」分割，左边的子组的熵会是 0.97（6 个「不好的」，4 个「好的」），而右边的熵会是 0（该组只包含 1 个对象），信息增益大约是  0.11。如果以这样的方式计算每种分区的信息增益，那么在使用所有特征构造一棵大决策树之前就可以选出每个数值特征的阈值。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 树的关键参数"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "理论上讲，我们可以构建一个决策树，直到每个叶节点只有一个实例，但这样做容易过拟合，导致其在新数据上的表现不佳。如果你这么做，在树的最深处，可能会存在由无关紧要的特征组成的分区，例如根据「客户裤子的颜色」这一特征进行分区，这是我们不希望发生。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "但在两种情况下，树可以被构建到最大深度（每个叶节点只有一个实例）："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 随机森林。它将构建为最大深度的单个树的响应进行平均（稍后我们将讨论为什么要这样做）。\n",
    "- 决策树修剪。在这种方法中，树首先被构造成最大深度。然后，从底部开始，基于交叉验证来比较有分区/无分区情形下树的质量情况，进而移除树的一些节点。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下图是过拟合的决策树给出的分界。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img width='400px;' src='https://doc.shiyanlou.com/courses/uid214893-20190505-1557025021753'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "常见的解决决策树过拟合的方法为："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 人工限制深度或叶节点的最少样本数。\n",
    "\n",
    "- 对树进行剪枝。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### scikit-learn 的 DecisionTreeClassifier 类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "sklearn.tree.DecisionTreeClassifier 类的主要参数为："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- max_depth 树的最大深度；\n",
    "- max_features 搜索最佳分区时的最大特征数（特征很多时，设置这个参数很有必要，因为基于所有特征搜索分区会很「昂贵」）；\n",
    "- min_samples_leaf 叶节点的最少样本数。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "树的参数需要根据输入数据设定，通常通过交叉验证可以确定参数范围，下文会具体讨论交叉验证。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 回归问题中的决策树"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "当对数值变量进行预测时，我们构造决策树的思路和分类问题时所用的思路是一样的，但衡量决策树好坏的质量标准改变了，现在它的质量标准如下："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$D = \\frac{1}{\\ell} \\sum\\limits_{i =1}^{\\ell} (y_i - \\frac{1}{\\ell} \\sum\\limits_{j=1}^{\\ell} y_j)^2, $$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "其中，$\\ell$ 是叶节点中的样本数，$y_i$ 是目标变量的值。简单来说，通过最小化方差，使每个叶子中的目标特征的值大致相等，以此来划分训练集的特征。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 示例"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "让我们基于以下函数生成一些带噪数据："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$f(x) = e^{-x ^ 2} + 1.5 * e^{-(x - 2) ^ 2}$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "接着在生成的数据上训练一颗决策树，并进行预测，调用 `plt` 方法画出结果示意图。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.tree import DecisionTreeRegressor\n",
    "\n",
    "n_train = 150\n",
    "n_test = 1000\n",
    "noise = 0.1\n",
    "\n",
    "\n",
    "def f(x):\n",
    "    x = x.ravel()\n",
    "    return np.exp(-(x ** 2)) + 1.5 * np.exp(-((x - 2) ** 2))\n",
    "\n",
    "\n",
    "def generate(n_samples, noise):\n",
    "    X = np.random.rand(n_samples) * 10 - 5\n",
    "    X = np.sort(X).ravel()\n",
    "    y = (\n",
    "        np.exp(-(X ** 2))\n",
    "        + 1.5 * np.exp(-((X - 2) ** 2))\n",
    "        + np.random.normal(0.0, noise, n_samples)\n",
    "    )\n",
    "    X = X.reshape((n_samples, 1))\n",
    "    return X, y\n",
    "\n",
    "\n",
    "X_train, y_train = generate(n_samples=n_train, noise=noise)\n",
    "X_test, y_test = generate(n_samples=n_test, noise=noise)\n",
    "\n",
    "\n",
    "reg_tree = DecisionTreeRegressor(max_depth=5, random_state=17)\n",
    "\n",
    "reg_tree.fit(X_train, y_train)\n",
    "reg_tree_pred = reg_tree.predict(X_test)\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(X_test, f(X_test), \"b\")\n",
    "plt.scatter(X_train, y_train, c=\"b\", s=20)\n",
    "plt.plot(X_test, reg_tree_pred, \"g\", lw=2)\n",
    "plt.xlim([-5, 5])\n",
    "plt.title(\n",
    "    \"Decision tree regressor, MSE = %.2f\"\n",
    "    % (np.sum((y_test - reg_tree_pred) ** 2) / n_test)\n",
    ")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图表明，决策树使用分段的常数函数逼近数据。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 最近邻方法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "最近邻方法（K 近邻或 k-NN）是另一个非常流行的分类方法。当然，也可以用于回归问题。和决策树类似，这是最容易理解的分类方法之一。这一方法遵循紧密性假说：如果样本间的距离能以足够好的方法衡量，那么相似的样本更可能属于同一分类。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "比如，根据最近邻方法，下图中的绿球将被分类为「蓝色」而不是「红色」，因为它与蓝球的距离更近。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src='https://doc.shiyanlou.com/courses/uid214893-20190505-1557025005732'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "再举一个例子，如果你不知道蓝牙耳机属于什么类别，你可以查找 5 个相似的耳机，如果其中 4 个标记为「配件」类，只有 1 个标记为「科技」类，那么根据最近邻方法，它属于「配件」类。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在最近邻方法中，为了对测试集中的每个样本进行分类，需要依次进行以下操作："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 计算训练集中每个样本之间的距离。\n",
    "- 从训练集中选取 k 个距离最近的样本。\n",
    "- 测试样本的类别将是它 k 个最近邻中最常见的分类。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在回归问题中应用最近邻方法很简单，仅需将上述步骤做一个小小的改动：第三步不返回分类，而是返回一个数字，即目标变量在邻居中的均值或中位数。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "这一方式的显著特点是它具有惰性：当需要对测试样本进行分类时，计算只在预测阶段进行。由于这种特点，最近邻方法事先并不基于训练样本创建模型，这与上文提到的决策树不同。决策树是基于训练集构建的，在预测阶段仅通过遍历决策树就可以实现快速地分类。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 最近邻方法的实际应用"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 在某些案例中，k-NN 可以作为一个模型的基线。\n",
    "- 在 Kaggle 竞赛中，k-NN 常常用于构建元特征（即 k-NN 的预测结果作为其他模型的输入），或用于堆叠/混合。\n",
    "- 最近邻方法还可以扩展到推荐系统等任务中。\n",
    "- 在大型数据集上，常常使用逼近方法搜索最近邻。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "k-NN 分类/回归的效果取决于一些参数："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 邻居数 k。\n",
    "- 样本之间的距离度量（常见的包括 Hamming，欧几里得，余弦和 Minkowski 距离）。注意，大部分距离要求数据在同一尺度下，例如「薪水」特征的数值在千级，「年龄」特征的数值却在百级，如果直接将他们丢进最近邻模型中，「年龄」特征就会受到比较大的影响。\n",
    "- 邻居的权重（每个邻居可能贡献不同的权重，例如，样本越远，权重越低）。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### scikit-learn 的 KNeighborsClassifier 类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`sklearn.neighbors.KNeighborsClassifier` 类的主要参数为："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- weights：可设为 uniform（所有权重相等），distance（权重和到测试样本的距离成反比），或任何其他用户自定义的函数。\n",
    "- algorithm（可选）：可设为 brute、ball_tree、KD_tree、auto。若设为 brute，通过训练集上的网格搜索来计算每个测试样本的最近邻；若设为 ball_tree 或 KD_tree，样本间的距离储存在树中，以加速寻找最近邻；若设为 auto，将基于训练集自动选择合适的寻找最近邻的方法。\n",
    "- leaf_size（可选）：若寻找最近邻的算法是 BallTree 或 KDTree，则切换为网格搜索所用的阈值。\n",
    "- metric：可设为 minkowski、manhattan、euclidean、chebyshev 或其他。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 选择模型参数和交叉验证"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "机器学习算法的主要任务是可以「泛化」未曾见过的数据。由于我们无法立刻得知模型在新数据上的表现（因为还不知道目标变量的真值），因此有必要牺牲一小部分数据，来验证模型的质量，即将一小部分数据作为留置集。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "通常采用下述两种方法之一来验证模型的质量："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 留置法。保留一小部分数据（一般是 20% 到 40%）作为留置集，在其余数据上训练模型（原数据集的 60%-80%），然后在留置集上验证模型的质量。\n",
    "- 交叉验证。最常见的情形是 k 折交叉验证，如下图所示。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img width='500px;' src='https://doc.shiyanlou.com/courses/uid214893-20190505-1557024973868'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在 k 折交叉验证中，模型在原数据集的 $K-1$ 个子集上进行训练（上图白色部分），然后在剩下的 1 个子集上验证表现，重复训练和验证的过程，每次使用不同的子集（上图橙色部分），总共进行 K 次，由此得到 K 个模型质量评估指数，通常用这些评估指数的求和平均数来衡量分类/回归模型的总体质量。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "相比留置法，交叉验证能更好地评估模型在新数据上的表现。然而，当你有大量数据时，交叉验证对机器计算能力的要求会变得很高。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "交叉验证是机器学习中非常重要的技术，同时也应用于统计学和经济学领域。它有助于我们进行超参数调优、模型比较、特征评估等其他重要操作。你可以点击 [<i class=\"fa fa-external-link-square\" aria-hidden=\"true\"> 这里</i>](https://sebastianraschka.com/blog/2016/model-evaluation-selection-part1.html)以了解更多关于交叉验证的信息 。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 应用样例"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 在客户离网率预测任务中使用决策树和最近邻方法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "首先读取数据至 DataFrame 并进行预处理。将 State 特征从 DateFrame 转移到单独的 Series 对象中。我们训练的第一个模型将不包括 State 特征，之后再考察 State 特征是否有用。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"../../data/telecom_churn.csv\")\n",
    "\n",
    "df[\"International plan\"] = pd.factorize(df[\"International plan\"])[0]\n",
    "df[\"Voice mail plan\"] = pd.factorize(df[\"Voice mail plan\"])[0]\n",
    "df[\"Churn\"] = df[\"Churn\"].astype(\"int\")\n",
    "states = df[\"State\"]\n",
    "y = df[\"Churn\"]\n",
    "df.drop([\"State\", \"Churn\"], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Account length</th>\n",
       "      <th>Area code</th>\n",
       "      <th>International plan</th>\n",
       "      <th>Voice mail plan</th>\n",
       "      <th>Number vmail messages</th>\n",
       "      <th>Total day minutes</th>\n",
       "      <th>Total day calls</th>\n",
       "      <th>Total day charge</th>\n",
       "      <th>Total eve minutes</th>\n",
       "      <th>Total eve calls</th>\n",
       "      <th>Total eve charge</th>\n",
       "      <th>Total night minutes</th>\n",
       "      <th>Total night calls</th>\n",
       "      <th>Total night charge</th>\n",
       "      <th>Total intl minutes</th>\n",
       "      <th>Total intl calls</th>\n",
       "      <th>Total intl charge</th>\n",
       "      <th>Customer service calls</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>128</td>\n",
       "      <td>415</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>25</td>\n",
       "      <td>265.1</td>\n",
       "      <td>110</td>\n",
       "      <td>45.07</td>\n",
       "      <td>197.4</td>\n",
       "      <td>99</td>\n",
       "      <td>16.78</td>\n",
       "      <td>244.7</td>\n",
       "      <td>91</td>\n",
       "      <td>11.01</td>\n",
       "      <td>10.0</td>\n",
       "      <td>3</td>\n",
       "      <td>2.70</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>107</td>\n",
       "      <td>415</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>26</td>\n",
       "      <td>161.6</td>\n",
       "      <td>123</td>\n",
       "      <td>27.47</td>\n",
       "      <td>195.5</td>\n",
       "      <td>103</td>\n",
       "      <td>16.62</td>\n",
       "      <td>254.4</td>\n",
       "      <td>103</td>\n",
       "      <td>11.45</td>\n",
       "      <td>13.7</td>\n",
       "      <td>3</td>\n",
       "      <td>3.70</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>137</td>\n",
       "      <td>415</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>243.4</td>\n",
       "      <td>114</td>\n",
       "      <td>41.38</td>\n",
       "      <td>121.2</td>\n",
       "      <td>110</td>\n",
       "      <td>10.30</td>\n",
       "      <td>162.6</td>\n",
       "      <td>104</td>\n",
       "      <td>7.32</td>\n",
       "      <td>12.2</td>\n",
       "      <td>5</td>\n",
       "      <td>3.29</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>84</td>\n",
       "      <td>408</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>299.4</td>\n",
       "      <td>71</td>\n",
       "      <td>50.90</td>\n",
       "      <td>61.9</td>\n",
       "      <td>88</td>\n",
       "      <td>5.26</td>\n",
       "      <td>196.9</td>\n",
       "      <td>89</td>\n",
       "      <td>8.86</td>\n",
       "      <td>6.6</td>\n",
       "      <td>7</td>\n",
       "      <td>1.78</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>75</td>\n",
       "      <td>415</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>166.7</td>\n",
       "      <td>113</td>\n",
       "      <td>28.34</td>\n",
       "      <td>148.3</td>\n",
       "      <td>122</td>\n",
       "      <td>12.61</td>\n",
       "      <td>186.9</td>\n",
       "      <td>121</td>\n",
       "      <td>8.41</td>\n",
       "      <td>10.1</td>\n",
       "      <td>3</td>\n",
       "      <td>2.73</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Account length           ...            Customer service calls\n",
       "0             128           ...                                 1\n",
       "1             107           ...                                 1\n",
       "2             137           ...                                 0\n",
       "3              84           ...                                 2\n",
       "4              75           ...                                 3\n",
       "\n",
       "[5 rows x 18 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "之后将数据集的 70% 划分为训练集（X_train,y_train），30% 划分为留置集（X_holdout,y_holdout）。留置集的数据在调优模型参数时不会被用到，在调优之后，用它评定所得模型的质量。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "接下来，训练 2 个模型：决策树和 k-NN。一开始，我们并不知道如何设置模型参数能使模型表现好，所以可以使用随机参数方法，假定树深（max_dept）为 5，近邻数量（n_neighbors）为 10。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',\n",
       "           metric_params=None, n_jobs=None, n_neighbors=10, p=2,\n",
       "           weights='uniform')"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import StratifiedKFold, train_test_split\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "X_train, X_holdout, y_train, y_holdout = train_test_split(\n",
    "    df.values, y, test_size=0.3, random_state=17\n",
    ")\n",
    "\n",
    "tree = DecisionTreeClassifier(max_depth=5, random_state=17)\n",
    "knn = KNeighborsClassifier(n_neighbors=10)\n",
    "\n",
    "tree.fit(X_train, y_train)\n",
    "knn.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用准确率（Accuracy）在留置集上评价模型预测的质量。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.94"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "tree_pred = tree.predict(X_holdout)\n",
    "accuracy_score(y_holdout, tree_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.881"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn_pred = knn.predict(X_holdout)\n",
    "accuracy_score(y_holdout, knn_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，决策树的准确率约为 94%，k-NN 的准确率约为 88%，于是仅使用我们假定的随机参数（即没有调参），决策树的表现更好。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "现在，使用交叉验证确定树的参数，对每次分割的 max_dept（最大深度 h）和 max_features（最大特征数）进行调优。`GridSearchCV()` 函数可以非常简单的实现交叉验证，下面程序对每一对 max_depth 和 max_features 的值使用 5 折验证计算模型的表现，接着选择参数的最佳组合。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 4 candidates, totalling 20 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 2 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  20 out of  20 | elapsed:    4.6s finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise-deprecating',\n",
       "       estimator=DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=5,\n",
       "            max_features=None, max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, presort=False, random_state=17,\n",
       "            splitter='best'),\n",
       "       fit_params=None, iid='warn', n_jobs=-1,\n",
       "       param_grid={'max_depth': range(5, 7), 'max_features': range(16, 18)},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score='warn',\n",
       "       scoring=None, verbose=True)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV, cross_val_score\n",
    "\n",
    "tree_params = {\"max_depth\": range(5, 7), \"max_features\": range(16, 18)}\n",
    "\n",
    "tree_grid = GridSearchCV(tree, tree_params, cv=5, n_jobs=-1, verbose=True)\n",
    "\n",
    "tree_grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "列出交叉验证得出的最佳参数和相应的训练集准确率均值。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'max_depth': 6, 'max_features': 17}"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_grid.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9425632233176168"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_grid.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.946"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(y_holdout, tree_grid.predict(X_holdout))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "绘制所得的决策树。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c7af706e0ea046a9a8af52905f2de9f3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Image(value=b'\\x89PNG\\r\\n\\x1a\\n\\x00\\x00\\x00\\rIHDR\\x00\\x00\\x11S\\x00\\x00\\x03\\x91\\x08\\x02\\x00\\x00\\x00n`\\x12~\\x00\\…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "dot_data = StringIO()\n",
    "export_graphviz(\n",
    "    tree_grid.best_estimator_, feature_names=df.columns, out_file=dot_data, filled=True\n",
    ")\n",
    "graph = pydotplus.graph_from_dot_data(dot_data.getvalue())\n",
    "Image(value=graph.create_png())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "现在，再次使用交叉验证对 k-NN 的 k 值（即邻居数）进行调优。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 2 candidates, totalling 10 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 2 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  10 out of  10 | elapsed:    1.8s finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'knn__n_neighbors': 7}, 0.8859837119588513)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "knn_pipe = Pipeline(\n",
    "    [(\"scaler\", StandardScaler()), (\"knn\", KNeighborsClassifier(n_jobs=-1))]\n",
    ")\n",
    "\n",
    "knn_params = {\"knn__n_neighbors\": range(6, 8)}\n",
    "\n",
    "knn_grid = GridSearchCV(knn_pipe, knn_params, cv=5, n_jobs=-1, verbose=True)\n",
    "\n",
    "knn_grid.fit(X_train, y_train)\n",
    "\n",
    "knn_grid.best_params_, knn_grid.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'knn__n_neighbors': 7}"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn_grid.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.89"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(y_holdout, knn_grid.predict(X_holdout))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，在 1-9 范围（range 不包括 10）内最优的 k 值为 7，其交叉验证的准确率约为 88.5%，调优后 k-NN 在留置集上的准确率约为 89%。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "综上所述，在这个任务里，决策树有着 94%/94.6%（留置法/交叉验证调优后）的准确率，k-NN 有着 88%/89%（留置法/交叉验证调优后）的准确率，显然决策树的表现更好。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用 `RandomForestClassifier()` 方法再训练一个随机森林（可以把它想象成一群互相协作的决策树），看看能否在这个任务上有更好的表现。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9494233119813256"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "forest = RandomForestClassifier(n_estimators=100, n_jobs=-1, random_state=17)\n",
    "np.mean(cross_val_score(forest, X_train, y_train, cv=5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 4 candidates, totalling 20 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 2 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  20 out of  20 | elapsed:    8.2s finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'max_depth': 9, 'max_features': 6}, 0.9511358765537934)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "forest_params = {\"max_depth\": range(8, 10), \"max_features\": range(5, 7)}\n",
    "\n",
    "forest_grid = GridSearchCV(forest, forest_params, cv=5, n_jobs=-1, verbose=True)\n",
    "\n",
    "forest_grid.fit(X_train, y_train)\n",
    "forest_grid.best_params_, forest_grid.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.953"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(y_holdout, forest_grid.predict(X_holdout))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，随机森林有着 95.3% 的准确率。不得不说，决策树在这个任务上的表现非常好，即使是训练时间长得多的随机森林也无法取得比它更好的表现。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 决策树的复杂情况"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "为了继续讨论决策树和 k-NN 的优劣，让我们考虑另外一个简单的分类任务，在这个任务中决策树的表现不错但得到的分类边界过于复杂。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "首先，在一个平面上创建一组具有 2 个分类的数据点，每个数据点是两个分类中的一个（红色表示 $x_1>x_2$，黄色表示 $x_1<x_2$），其实用一条直线 $x_1 = x_2$ 就可以完成它们的分类，那么决策树会这么做吗？"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f536dfb57b8>"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAEBCAYAAAB2RW6SAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzsnXV4k1cbh++4tVAKbaFFSpHgMKxluLsNdzbcGTJkDBhjOAOGFx3uOtxhaHEbwV0qUE3StEm+P1JKkwptaBn0631du3b1cN7nPSdv8rxHfud5BGazmQwyyCCDDNIvwv+6ARlkkEEGGaQtGY4+gwwyyCCdk+HoM8gggwzSORmOPoMMMsggnZPh6DPIIIMM0jkZjj6DDDLIIJ2T4egzyCCDDNI5GY4+gwwyyCCdk+HoM8gggwzSORmOPoMMMsggnZPh6DPIIIMM0jni5FRSq9U7gLyACQgHBmg0mqtqtbog8BeQFQgCOms0mnvJvLcMKAe8AowpbXgGGWSQwf8pIiAH4AdEJueCZDl6oItGowkBUKvVTYHlQGlgETBfo9GsUavVHYHFQI1k2iwHnEpm3QwyyCCDDKypDPyTnIrJcvTvnXwMmQGTWq12xeLsa8eUrwfmqdVqF41GE5AMs68A3r2LwGRKeQTNrFkdCAoKT/F1XyrprT+Q/vqU3voD6a9P6a0/EL9PQqGALFlUEONDk0NyR/So1eqlQB1AANQDcgEvNBqNEUCj0RjVavXLmPLkOHojgMlktsvRv782PZHe+gPpr0/prT+Q/vqU3voDifYp2UveyXb0Go2mO4Bare4ETAd+Se61SZE1q4Pd17q4OKZGE74Y0lt/IP31Kb31B9Jfn9Jbf+DT+ySwJ/GIWq3WAZ6ABsgaM5oXYdmQLZDMpRtP4FFQULhdb2AXF0cCAsJSfN2XSnrrD6S/PqW3/kD661N66w/E75NQKHg/QM4LPE6OjY/KK9VqtYNarc4V5+/GwFvAH7gKtIv5p3bAlWQ6+QwyyCCDDD4TyVm6UQGb1Wq1Csua0FugsUajMavV6t7AX2q1eizwDuicdk0Fo9HIgAG9ObVnFwLAp259Fi5chkgkstum2Wzm8OED7Nm8AYFASKPWbalRozYCgeCTbB44sI99WzYiFItp0qY91arVSNDmy5cvWL1iKS8faMhbtCQdu3QjW7Zs8eqZTCb279/L/q2bEEulNG3bgSpVqiVoc8+e3YwbOZTIt0Fk8sjJ3EXLKF26rN39AXj27Cmrly/h2T0NRcv70KFTV7Jkcf4km2nBqVMnmDJ+DMGvXpK/dFkmTZ2Bh0euePWioqLYtWs7R3ftQJUpM606d6VcOe//oMWfn8jISLZt28ypfX+TPZcHzdp0okSJUp9kU6/Xs2XLRs4c3IeTiyvtunanePESqdTi1EOr1bJp0zrOHzmEi7sH7b/vQaFChT/JZnh4OJs2rcPv6BFcc+Wiw/c9KFhQnUotTh3sWrpJJTxJ4dLNN/lzIQ4NYSiWqchMQKdy4Pqjl3Y3YsTAvpzftZ2+2gjMwHylisotWvH7zD/tsmc2mxnSpzvXD+yld0QEUcB8pZI67ToxbvJ0q7o3blynbdN6tDEY8DYYOCaTs1+hYPuBY+TN62Vlc2CPLtw5fIhe2ggigXlKJY27dGP0r79b2Zw+fTLzpk/me6ACsBfYCcxZspKmTb+zq08XL16gc6umdDJEUTrKwEG5nOMODuw6dBIPj5yJXve5p9GLF89n8i+j6AOUBLYBR4RCdh89Q5EiRWLrRUdH07llEyKuXuEHbQRBAgFz5XL6jPiZnn0HJmo/PSwL6PV62jSqg+L+XTpqtbwUCpkrk/Hz5Bm0bd/JLptarZZWDWrh9Pgh7bVanopEzJdKGT9jDi1btU3lHiRNUs8oLCyU5nWqkevlS1rptDwQiVgklTJ1vi+NGjW1634hIcE0q10NrzevaKnTcVckYrFUyizfldStW/9TuhJLaizdfDWOft682cybMJZHQKaYsnDAC+g8dAQjRvyc4gZcuXKJHs0ackun5f2WcChQWKFgzZ7DFCtWPMU2L1w4z4DWTbmh1aKKKQsGCskVbDl0ArW6UGzdlnWr0fHKZXrEuf53oZDLdeqxaNWG2LLTp08xokNrrmojUMSUBQGF5XJ2Hz+Dl1f+2Lpqt8zMNputplZjgcVyOTef+qe4PwD1KpVj+F0NcX+yo0QiXjRvycwFSxK97nM6RpPJRCGPrKwyGmkcp7wfcEZdiIOnLsSW7dixlWWD+/OPNiJ2SvsUKCmTc/bqv2TNmjXBe6QHR79y5VIOjxvDPp02dt32X6CiQsnl2w9QqVRJXZ4gS3wXcPb3X9ml0/F+fnkDqK5Scfn2QxQKRVKXpypJPaM5f0zj3qzpbIiMjG3neaB5Zicu3n6ARCJJ8f2mT5nIq/lzWBXH5j9Ae+esXLh1/5NWG97zWdbovxTWrVtNez44eQAHLGtFmzdvSPiij3Ds2BFaR+qJq/vJBLSMiuLo0cN22Tx65CDtdDri/lycgO9MRo4d+2DTYDBw9tpVuthc381k4uiJY9Y2D+2nvU5L3J9LVqCJ2dKH9zx58oQwszl20+Q93YEwvd6u/oSEBHP30UNa2bbTaOTo4YN22UwLrl27gtFopJFNeU/gyb27VmXH/t7F93GcPEBuoJJEwunTJ9O4pf8tx3buoFscJw9QGCgiFnPpkp9dNo/v2sEPcZw8QHHASyjk6tXLn9Da1OXY7h10j+OQAbyBLMZo/v33ll02j+/eGc9mJUAWqUejufMJrU1dvhpHr1KpeJ1A+euYf7MHBwcHAqXSeOWBEgmOjvbJmRwzZSYwgZFBgFiMo+OH15RIJEImFvPOth6QSaGMb1Mc32agWISDw4d2Ojo6IgBCbOuRAh2tDVKpDDNgO0YKBByVygSu+G9wds5KFKC1KQ8AJDajKgcnJwKE8b/6gWD1eaZHHJ2c4h1yMQOBJqPd33lVpswEJmAzyGj6oj5PxwTaaQTeRhvtbqdDpkzxbEYBoUb7P8+04Ktx9JMnz+Bv4EycsgtY1mEnTpxql82mTVuwUyDgQpyyM8A+oHHjZnbZbN68BVuEIuKOY04CR4GGDT8sKohEIlo0a8FPMhnRMWV6YJRcTqtOXa1ttmjNOpGIa3HKjmKZIjZo8GEM6+zsjLNCwTA+nKTQAUMAl5zxNySTg0KhoEGdeoySSGNtaoGfFQpafd/dLptpQZ48nrhlycIvWAIygeXlNAKoUK+hVd3WnboyXybjQZyyLcBzqZTKlat+lvb+V7T+vjszlEpexPxtBpYIBAiyOFOqVGm7bLbt3pOpSmXsQMwMzBcIULq52bX8mVa07t6LiUplrGM2AzOEQnJ75cPLK59dNtv06M0EpZK3cWxOFYkoWKgwuXLlToVWpw5fjaMvW7Y8jVq2oSZQFkugnKpA7Sbf2f3jdHNz40/fldRXqqjm6EhVR0eaqBxYsHx1gsqX5ODu7sHMhUuopVRSw9GRyo6OtHBwYNHKdTg5ZbGqO27KDF58U4a8CiXNM2Uij1yOpEp1Bg8baVUvTx5Pps1dSDWFgpqOjlRydKSdYyaWrN5oNUsA2HzgGDuFQnJgOcacHbgtk3Hg2Bns5fdZ87hZrDj5lEqaOTriKZfjVqc+ffoNsttmWrB290HWyOTkwtL3HIDBMy8LFy+3qley5DcMGfsbZWQy6jk6Us7BkR+zZmPlpu12rdN+TVSpUo1Og4ZSVCajgaMjpR0dmZY9O8s2brdbaVajRm1a9x1IYZmMho6OlHRwYK67B0vXb/kk9Vpq06hRU+p07UYBmYzGjo4UUzmwKlduFqxab7fN5s1bUrVjF/LH2CyiUrEpjydzV6xNxZZ/Ol/NZiyATqdj+pSJrP5rOQIBtOvQhRGjx6L8xCUErVbL6dMnEQgEVKxYJdHNo3PnzrJw+iQe3LuLunBR+o0YnahsMSIigtOnTyISiahYsQpyuTzR+9+6dZOgoJe4u+clf/4CidYLDw/n9OlTSCRiKlasgkwmS7Tun3/O4uLFC9Sv35B27TomWi+5zJ07i2Uzp6HVacmcxZkxU2d+VMXzX2xemkwmNm1aj0Zzh7p16+Pj822idYOD33H27BlUKhXfflsJsTjhBa4DB/axbPZ03rx+Ramy3gwY8XOCzykgIIB5f0zl+P69ODg40KZ7bzp26oowgWWi/5rAwEDOnz9L3rweqNUlE9w0NJvN7Ny5jVXzZhMQEIB35aoM/Gk0uXPnSdCmv78/fn7ncXZ2xtu7QoL9NpvNbNu2mdXz5xAUFMS31WowYPgociYw43z+/Blzp0/mzPGjZM2alU79BvHdd63ivTzMZjObN29g7cK5hIQE41O1BgOHj8Ld3SPBdr5+/YqLF/1wcXGlfHnvBF9GZrOZDevXsG7xfEJCQqhcqw4Dho0ke/YcCdp8+fIFly9fwtXVjXLlyidqc+3av9iwYRFhYaFUqVKX/v1H4ubmlqDN9/xfqW7MZjNtG9fF4fpVftLrEQLTZXICixRhy76jaf5jOnr0EAO/78jvOh0VgBPAOIWCpeu38u23lT7Z/pes6Bgz+ie2LF3EbCxR7A5giX8xc+ESWrRok+h1X3KfksuqFUuZN34MU3VaigA7hULmKJXsPHjCytmHhoZQt7I3dQMD6BYVRQAwVqmkePOWTJ417z9r/8dI6hnNnTWdzXNmMlmrxQvYKBKxwsGRfSfOJupEP8YfUyaye9E8Jmm15AHWi0SsdszE/pPnrJzo69evqFfFh85hobQ1GnkCjFYqady7P0NGjrGyOfW3sRxa5svvWi25gDViMRsyZ2b/ifO4urra1c7fxozknzUrmajV4g6sEovZ6pSFA6cuJKrK+hjjxg3Bz28dv/2mxc0NVqyQ8PffWTlw4Hy82X5c/q8c/cmTxxnbpR3XIyJ4P/YwAaVVDoxcupKaNeukYVOhToXS/PrgPk3ilK0H5pcoyfbDnx5t+Ut1itHR0RTyyMrfZjOV45TPA2Y4ZcHv7pNEr/1S+5RcDAYDpQvn5UhYGMXilP8mFPJvk+bM8V0RW7ZowVxuTJ3IJp0utiwE8JLJOHz2coIj1i+BxJ5ReHgYZYrk54peR9zx+xCxhMguPzDe5kxIcggJCaZc8YLc1OuJ+5roL5Ei6t6TMb9Oii37bexozMt8mRtliC17DhSXy/G7cZfMmZ0AePs2CO+ShdBERhJ3XNxLKsWxzwBG/Dwuxe309/enYumi3DdEEncB93uZnByDhvCjzdJqcnj9+hVVqpTg4cNIssTx6R06yMmXbyQDBgxJ9Nr/K3nllSuXaKDXE3eCKQQaRYRz+fKlNL23yWTi2oP7NLQpbwxcvn07Te/9X/PkyWOizGZs5yxNgaBgW81Q+uL582cojCYrJw/Q1GTi6oVzVmVXT5+kSRwnD5Z43hWlUq5du5q2DU0D7t7V4CmRYLtI0zQ6istn7BvY3Llzh4ISKbZzgaZRBq6etg6rfvX0KZrEcfIAOYGCEikajSa27PbtWxSXybBd/GhiMHD1tH3tvHnzOmXkMmx36ZpE6rly6oRdNq9du4q3t9TKyQM0barn6lX7bKaEr8bRu7t7cCuBde5bSlWSpzNTA6FQSPbMTvxre2/A3c5p3NeCm1t2zFhGU3G5BSgTkKamJ7Jly0awMTpWUfGeW4C7zQjdPa8XN23W+E3AbaMJDw/7ljn+S3LkcOepwRBPrnoTcE9kjf7jNnPwKMoQLyXSTYEAd09PqzL3PJ7ctFnn1gMPDQZy5PiwxOPu7s49QxTWrwS4KRSSI48n9uDhkZM7UVGxarhYmyIROfLmtcumu7sH//5rxGgTWPjmTRHu7l4JX5SKfDWOvmHDJlyXylggEBANRANLgPNSCU2a2CeFTAnd+vSnh0IZ6/AeA30USroPTHzKlR5wcHCgRLESdATexJRpgN5Aww62x73SF5kyZaZpk+Z0l8sJiim7BoxSKOk2eKhV3Y4/9GCpRMI+LBI7LTBSLMbZ05OSJb/5vA1PBXLkcKdS5Sr0kcliz2WcAyYqFHzf/0e7bObOnYey5X3oJ5USGlP2DzBVLqerTeiJrv0GMlUu53TM36FAf6mU8j4VrGSLXl75KV7qGwZJpIRh+exPAH/IZHzfd4Bd7VSrC5G/SDF+lEgJj7F5FJgrldGlZz+7bBYvXoKcOQsyfLgErRbMZjh4EBYulNGpU2+7bKaEr8bRK5VKNu4+wF+Fi+AqleImk+FbqDAbdu5PlUMZUVFRREfbvsM/0H/wMCp270kJhYI8ShVllErq9R3AD9172W3zS+Fj7dz090EiChXGE3ifVqxE3fpMnTozSZtG2+FLApjNZvR6Pam5V5QSm+Hh4RgMtuPBD0yc+SeZGjcnn0xGbqWSepky8+PEKdSqVdeqnpdXfhat2sBgdw9yyOW4S6XcqFCRFZt2flESw5Qwy3cF+tp1yS2VklOhpLVzVn6bswBvb59ErwkNDU3yuzRvxRqCa9Ymp0SCu1xOh2zZmDzfN556rUyZckyat5j22bLhLpeTUyIhuGZt5i5fHc/mwtUbeF2tBjmlUnIqlXR1cWXG4hUUL17S7r4vWbeZp5Wr4iGV4qFQ0MPNjdlLV1KkSNFEr4mMjMRkMiX678uWbUejqYSbm5js2WX065eDefPWfJYAaF+NowfLyc8c7h5ojUYioqPJnt2dTJkyffzCJHjx4jnd2n6HV25X8uZyoUf7lrx6FT9ImlAoZMQvE7jy7yO2HD/D1TuP+XHEzwn+iJ8+fULXVk3Jm9sVr9yu9OrUhjdvEjrX+9/y8OF9OjZrQN7cruTL7Ub/HzoRFBQUr55CoaBxy7Y4OmbiLZAjew6aJqK2uXfvLu0b18UzlwuZFAoG9ejKu3e2ix+WfY+5s6ZTokAu8nlmp1qZYuzZs+uT+mMymZg9fTLF81tsVi9XnP379iRYd9euHZTyzEEBL3fy5sxGtbLFefnyRbx6EomE7LlyIRKLeaXTkdXJCdccCUvsnJ2dcXN1JSgykmizmezuHiiVny/OS2rj4ODIvOVruHTrPjtOnuPCrfs0bdYiwbp//bWcYrlcUOfPSV53Z+pV9iY4ODhePbPZ8vswREXxRq9HFxrKmzdvErAILi6uuGbNRqDBgFAkIkfOXEil8SXFmTM74bt2Exdv3uP0rVucu3GXevUafFLfs2RxZumGrfjduMvuUxc4c01D7dr1Eqx74sQx6tYtg5eXG0WKeDB58hiioqLi1TOZTDx79gyDIZqgoEjCw4MJCrI9V5s2fDWO3mAw0LJ+TYodO4K/0UiA0UiZU8f5rl4N9HbGcdHr9bSoX5PSJ44RYDTibzRS7NgRWtavmegoT6lU4umZN1FdvFarpUX9mnx76gRvjUZeR0dT8MghWjWs/UWN7kNDQ2hRvxa1z57mndHI8+gosh/YS/sm9eKNSubOms7uP6ZyNCyUKGDp61eMG9SHQ4f2W9V79+4tLRvUotGFc4SYTDyNisJp7246Nm8Yb3Q9a+rvHJgzk2OhoUSaTMx9/ozRfXtYxe5JKdMmjuP4vDmcCLPYnPX0KT/1/oF//rGOX3Pz5nUGdu/ML9oIIrDsP5R5+oT6FcvFszlp3GjOLJzL2YgIIs1mpj99wuBunTl3zvoA2suXL2jbtD5drl4hwmzmflQUxh1b6dmhtd39+VLInNmJPHk8Ew3QdeTIIcYNH8zsyEj0wD0gh+ZfGlSK/3k2rOKN862b3AVLFFaDgYmjhrHP5oV8586/fN+mOcM1d9CZTFzX63my+i9+6p/4DNrJKQuenp6pKrXOksWZ3LnzJGrz6tXL9OnTlnHj7qHXm/Dzi+DmzSWMHRt/eat+/TIUKnSfx49Bp4PJk3UMH94z3ncpLfhqHP2BA3txDw1hotFIJsARGG80kj883O6R4J49u8gXFsZ4oxFHLAHNJhqNuIeGsH9/wiPBj7Fr13aKRkTws8mECovyYkp0NM5vgzh06IBdNtOCzZs2UFGvZ5jZjBJwBmZHRWF8+cLKMUZHR7No3mzWarUUw5IwuBowR6djwRTrEMkb16+lZmQkg8xmFFgCr82PiiL8yWOrL7PBYGDJ4vms12opHGOzJvCHTseCqdY2k4tOp2PFMl826LQUirFZB5iq07Fw2iSrur/+PIJ2WPYZZFiWo1YApohwtm7dGFsvPDyc1atWsFGnoyCWH0t94HedjkUzpljZXLV8CW0NBroBUiwnkpdHRvLv1cvcvm1fwKyvhcm/jGQIlsxDYixJozcBr/zf4Od3Prbe9etXefbyBVuwBJETA62xhKmYNnaUlc2lc2fxY2QkrQFRTP21eh379+/h9etk58ROc3x9Z/Dzz3qaNgWhELy8YMMGHVu3brSayR4+fJDQ0GDWrIEcOUAiga5doUcPmDgx5XLNlPLVOPoHD+7jbSNfA/COCOfBg/t22Xz48AHlI+JnjPfW6Xj48EECVyTD5v17+Ggj4pWX10fabTMteKj5Fx+dtaZCAHgbjVafZ2hoCNEGA7ariN7AgyePrMoe3bmNt976GQmA8mazVd/fvXuL2GTCVmtQHnj0+KFd/QkMDEApEGCrVi8P8T731w8fxJOLirCE1bh06WJsmb//a7KIRNgu1Hhjec5xefzvLbxtZoFioIxY8kU997Tg3cuX2J4/VgJq4OLFDxExL126hBdgmyW6AhBss7T56M6/eNvMLB2BgjIZT54kfnbjc/Pw4R18fKxnq87OkDOnlOfPP2jV/PwuULIk2ArVKlaEwMC0789X4+gLFy7KMbmCuB+pGTimcqBw4cQ3SJKiUKEiHHdwiG9TrrDbZuGixTiqim/zuExG4cJFErvss1O4RCmO2oSOMAEnREKrvmfO7IRSqcRWCX4MKFywkFWZutQ3HLOxacSigojbd2fnrCCRYDvOPQ4ULmTfZ+Tq6oZeIOCuTfkxoLDNBlruokWxnVsZgNNA5crVYsty5PAg2GTmkU3dowIBhYpaB+tSf1OGYzYhKfTAuSgDhQt/WgajLx1XT09sg3qHYIlzX7lyldiyihUrcR/iyVUPA9lyWwcAU5f6hmM2ctW3wJ3ISPLly8+XQqFCpTh61NqNvnoFz59H4RlHMlqlSlUuXQKtjV714EHIkaNgmrfzq3H0tWrVIdrdg14SKY+wyBv7SSSEuLravfFSr14DQlxc6SeR8Bh4BPSSSIlyd6dWLftO2jZo0JjXWbMySCzhKfAA6CaVIs6dh2rVathlMy1o/l0r/s2cmZ/EYp4Dd4GOMhmuBQtZqSpEIhGDR4yhjULJUSxJVDYBQ+UKBo0Zb2Wzdet2XHZwZLRIxEssMsz2MhlexUvyzTdlYutJJBIGDRtJK6WS41hyUK4DRioUDBid8pOMADKZjP6Dh9FCqeRkjM3VwFiFgv42R+Yn/D6d3QIB47GEub4FNAMyO2elfv0Px+IUCgV9+w+muULJP1gczUrgN7mc/iNGW9ns2KUb+xRKfhcKeYNFb95CLqdi1Rrky5d4/KL0wK/TZrEImI0lLPRloB6QP68XxYp9SCdYsKCawgUKUg+4hCUs9NyY/8ZOtlZw9eg/GF+ZjLkx9S4BTRUK2rRpb3fAwbSgT5+f+OMPOb6+EBQE585B06ZKvv++p1XAwYoVK+PunoNGjeD6dXjzBqZOhXXr4LffElevpRZfjaMXiURs3HMQU5t2eDs4Ut7REX3LNmzZdyTRiIOBgYFcvHiBwMCEd7YlEglb9h1B37IN5Rwc8HZwxNSmHZv2HLI7M4xMJmPb/mMEN/uOEjIZ5eUKRG3as2H3/s8a3OrmzeusWbOSR48SXgpRqVRsP3CcZw2bUFQu51uVA04durBq6+54SqKu3XoycPoseuVwx10kZrq6MIvWbooX48fBwZEdB4/zqGETiimV1MySBdcu3VieQGTE7r3702fSdPrnzkMemYxFpb5hyfqtlC+fcN7W97k+9+/fm6iEre/AIfzw2xT65MyFp0zGstJlWL5xRzzpXr58+Vm7fQ9r3bKTF/ARCAguXZbDF67Fszlw2Ag6jp9IV7fs5JJKWVaqNKu27IqXY9XFxYVt+49yqVYdCspk1M2cmSK9+zN32aoE2/pfExwczPr1qzl06NBH6z558phLl/yIiIi/JAlQvrwPC/9az+wsWcgFVBcIUFWtzr5TF+LV3X3sDFlq1KSmUEhOYKaTE38uXRUvAm3evF5s/vsQu3y+xVMioZlTFmoOHcGvU/9ItJ0PHz7gwoUL6BJY4k0r1OpCrF+/lw0bfMidW0K7di40a/YLI0f+Fq/uwYNXiI6uQMWKkCcPLFzozIoV26xehmnFVxPrxpak4qhER0czZtggtm3dRH6pjPuGSJp/Z8kDm1iEwtRk2TJfJo0eTmazGSOgFYqYNHchrZLIn5lacWGCg4NpXNWHZ69ekgfLLKVE0WLsOHQyXt+PHz/K0N7dcDBEojWZULm4suCv9fG0wu/evaVhFR9evXlN7hib35T8hu0HjiX58kqtPk2ePBHfWdNxxUwYYJJIWLhmI9Wr1/pk2x8jLCyUgd06c/7cGTylUu4bDPQfNJQBQ36K9/LatXMbo4cMxM1sJsgYTY48nixavZE8dp7QTCuG/difzWtX4YElJaVUoWD1zv2UKmV9sCsoKIh+Xdpx8/pVPCQSHkcbGTFmPD/0SPsDPgD9en7P3zu2khPwB5QqFev3HLHK/wvw5s1r+nRqw707d3CXSnhqNDJ24lTadeicoN3Uxtf3T2bO/J28ecU8exZFqVJlmDdvLVmyOFvV27NnN4MHd0QoNOPsDC9fgo9PNTZuTFpM8n8V6yYlzJ4+mcfbt/AoMhK/sFAeRUbydMdWZtsoJdKCu3c1TBg1jA1mM8+AF4Cvycjwfj0JCLDN7ZP6tGlch3yvXvIGuI0lF6r+1k16dbP+0r98+YLeXdqz8m0Qt8PDeaTVMvzJYzo0b0hkpPUh9Zb1alL0zetYm4+B0GtX6Nsn7ROPnDhxDN9Z0ziAmYdYTudOiIqie7uWSR50Si2G9+uJ25l/eKbX4xcayjW9ni1zZ7Fr13arerdv32LUwD7sCQvlengYT3U62tzV0KVlk1Q9DPaprF27il1fP1oMAAAgAElEQVRrV3EeiwzSH+it09G+cZ14M6X+XdtT8vIlnun1XA4L47xOy8Lfx3P8+NE0b+fChfM4tmMrV7AsKwYAHSIiaNegZry6Pdq1pOqN6zzX67gcGsqJiAimjh7O+fPn4tVNbQ4d2s/y5ZO4eFHHpUthPHumR632Y/Dgrlb1jEYjgwZ1YPhwMwEBcP8+nD4Nly4dZ/bsGWneznTp6Fcv92WuTodTzN9OwJ86HauXLU7ze0+e/BtNgfe7BgIsErJvgalTJ6bpvbVaLbc1d/Dlg7LBBVgA/HNwn1XdzRvX0doYzftdAwHQFSgQZeBwnFywoaGh3H30gMUQmwfXDcu66sndO9KqK7HMnDSBXkDFmL+FWBJ+5zaZWLBgbpre+927txw5eoTZBgPvT03kBn7Xalk9f45V3XUrltDXYOD9IpEIGGYyQWAgfn7xlzD+K3xnTuNnLDldwaIMGgsIIyPZHed5Pn36hOvXrjAlOor3QpECwC9aLasXpu3nDrBq7iwmQKzaSwJMArTaCE7Eyams0dzh+YP7/Go0xqbLLAr8pNezZvH8tG/nqjn8+qsWrxgJmUwG06YZuHDhnJUMdPr0STg6wujR8H5V+JtvYPhwWLFiVpq3M106+oCwMGxDD+UF/ENDE6qeqgS+eU1Ce+gFgNev0/Z0bGhoCCaIJwfMC+htwhEEvn6FVwIj4rxGIwEB/rF/BwUFIsCiNbe1qfsMB8BCA/1JSGPhBbx48SxN7x0cHEwmsQjbABt5Id7sLOjVS7xsPmMB4CkUWH2e/zW6kOB4slYhlhfYkyePY8uCggJxl0iwDVuXFwhI4+8xgD4iPF47JVi+248ff9BBBQYGkFsixnZHLa/ZTOBn0NsHBvpjG+dMoQAXF7HV3uDDhw/IkwdsD9LnywdGY9rvKaRLR/9tyVJssinbBFQslfbBpWrXa8AaLAmC36MDtgKNGjVJ+KJUwtXVjUxiMX/blG8A3Jyt1wt9qlRns0pFXNcUDuw1Y7XJmiePJwqRKJ4ccSOQ3c6kDimhXI3a/AVWctVgLEGmWrRI21OnuXPnwSyTc96mfKNYjE/V6lZl3rXqsFGhtGqnP3DGYKBs2fJp2s6UUNDbG9vt4edYVEJxM4YVKlSE5yYTd2zqbpBKqVCzdhq3EvKWKIVtMr57WJYNGzT4kHu5RImS3DYY4i1Ub5TJ8bFTOZcSKlSoxYYN1mKQ69fh7VsBBQp8GPJ169aby5ct0sv3mM3w11/g6poR68YuRk+ewVClkgkiEceACSIRQ5VKRk1KebKElNK370AMmTNTDdiJJen0t0Cm7Dlo27ZDmt5bKBTSb/Q4OgBTsWjIR2LJBvXrLOtpbN269VEULkpDhYLdWBx3VaWSek2aWQVZEgqF9Bo+itbAjBibPwETgN9mL0jT/gD88ssEHsjl1Af2YpFhlgcKFS6SZJrA1EAkEjFu6kyaKRT8KRBwBBgokbDawZEBNsknWrduzwsPD1rL5ezHIu2sqlTSrWffj6aK+5xMm/Enx0Ui2mDJFLYC8AGqVK1htWmsUCgYPX4idRRKFmPRuveQSjnklIUeNpEm04Lps+ezUyikM3AIWApUAuo2bIyLi0tsPUfHTAwd8TM1lEqWAQeBLjIZF7Nlo2u3nmnezl69hrBzZ2Z695Zy+DAsWAANGigZM2aKVapPb28fcuXKRfnyFue+fz+0bAlnzsCyZWmfX/ajqhu1Wp0Vy/c2H5ZzJfeAXhqNJkCtVpuBG1jO2gB00mg0N5J5b0/SSHUDlk3RJX/+wb1bNyhQtDg9Bg75LFHiwHJ0fujQAfgd2I9AIKBSk+ZMnz4LaRLx21MzG9PWrZuZNGYEoSHBuLjlYNrchVSqVCVePb1ez5pVKziwZSNSmYzmXX7gu+9aJaik2bp1I7N/G09YUBCuefIwYcacjzraj/XJaDRy/PgRnj59SqlS31CqVOkEg8QFBQUyeGAfbv5zErFEQt22HZkwYVKC7YyOjubYscM8f/6c0qXLpEqI4DNnTjNzwi+89X9NCe8KjBo3McHcoeHhYfwxcxr7d2xF5eBA3yE/0axZi0+OXrl58waWLfPFycmJGTPmJJqtymCw7K/4+7+hfHmfRCMtPnv2hL49vufx9WtI5DJadu/D6NG/JFj31KkTrFo4l8DXr/CuXovufQZ8Nh37gwf3GTawD49uXEOmVNGx30AGDEg4RPLRo4dZu3g+YcFvKV+jDt169omnekkpWq2WQ4f2ExwcTOXKVfDySviglr+/P0uXzsXP7yguLh507Tow0fSiXbt24OzZPYCJTJk8WL16K4U+ckjws6QSVKvVzkAJjUZzPObv6YCzRqPpFuPoHTUaTfw4Ah/HkzR09F8bqdUff39/2jWpi9j/DWWjjRwTCslVrDgrNu345CTqKSWpPr18+YK2jeuS6d1bSkQbOSwUUKhMeZas25xk0vOkePbsKW0b1yVraAjFoo0cFAgo4fMti1ZvSPIlmxSPHz+ibZN6ZA8Po6jRyD6gbOVqzF+xJt75jam/jWOl70IaCoUECIX4CWD5ui12zzyMRiPexQvwNjCQxsBL4DzQvf9gxo6dYFX33r27tGtaH0+9jgJGI3vNULVufWYtWmZ1JsRsNvPr6OFsXruKhgIBr8RirgqErNq8w+pQ29dKav2OLl++SJcuzSlZ0oi7u5E9e8y0bt2FsWOn2f3i9vM7T9eu31GmjBk3N4vNDh26M3r070na/CzySo1G8/a9k4/hHMTLMJbBF8KYIf2p+/QJl8LD8dXruKONINu1K8yaZl+wsLTip749aPvyBefDw1mi13FXq0Xkd46F82bbbXNorx/o+uY1Z2Ns3tNpMZw9zZJPUF8M6dGVPv5vOB0ejq9Ox32djtBTx1mxzNeq3qlTJ9i+bDF3IvX8pdOyNyKcNeHh9OzUxm4Z6KBBfTEFBvIEy5LVcWAzsHzebMLDP4ytzGYz/bu2Z2RQICfCw1mq03Ffr+PJof1sWL/GyuaBA/s4tn4Nd/V6Vuh07A8LY2FoCD07tklW/oD/B4xGIz17tmHhwhAOHAhnxQodd+/qOXp0TbyIrcklOjqanj3bsGJFGPv2hbNypY47d/Ts3bv8kyK2JpcUrdGr1Woh0AeIq/A/rlarr6rV6slqtdq+oVgGqUJkZCQHjh5mTHQ078cHImBcZCTbN6z7L5tmRUhIMGf8zvNTHMciAX7R69m+5i+7bAYFBXHl+lWGxrEpBcbodWxfvdIum2/evOb2ndsMiqMvlwE/63RsX7XCqu72davpr9Va5RmtC+Q1mThtZ+7S4zu3MwpLZNH3NMKikJk48UOoiEePHvL6+TN6xZmdK4GRWi3b/1pu3c41Kxmi1cZKjwGaA5l0WquAbv/PXLzoh5OTnubNP5Q5OcGPP0awbdtKu2yeO3eG7NmjaBgn8XTWrDB4sJbt2+37zqeElB4TnYtFnDEv5u/cGo3mmVqtzoRlHf8XYExiFydEzBTELlxcPj2z1JfEp/ZHqxVhxuKM4qIAoozR/8nnldA9BYJIhAIBtoErPqWd0dHhiIXCeF9oJRBtp02t9i0SoTCedE8BGE3WNgXm6NhzBlb3FwhQKsX2ffZmEwkttikAo9EQazMgQIpCJMJ28m+pF2XTTmOCNlVCIUqlKF38pj61DyqVGJUq/lKKZeXTvu+SUpmwTYUCBALjR21+ap+S7ejVavUMLHLwxhqNxgSg0Wiexfw/VK1WLwVSnEA1Y43eQmr1x6dUaXwv+TEwzujuT7GYOvUafvbPK/E+yVDny8+qf2/zQ0yJGZgrkVKnUVO72ikWO5A7Z27W379Lxzg2/5RKqd24uV02FYosZHN1Y8uTx7wXcpqBuTIZNW1sVmvYjPn7D9BFGxH7or0BXI6KomjRMnbdv5BPRWadPEZ7iNWzX8ESFXLZoJ9ibWbN6gEqB/aFh8ce1DMB8+RyajZrad3Oxt+x4PRpWmm1sS/aC8ADk5n8+Yt99b+p1Pgd5c9fjHv3TPj5QbmY3ClRUbBwoZK2bVvY9ywLleLmzWiuXLEclAIwGGDRIhU//JD09zOJNfpkk6xYN2q1ehKWsNENNRqNNqYsC6DXaDQ6tVotBpYBbzUaTXIzB3uSsRkbS2r15969u7RsUIuqhkjK6XQcUqp4mCUL2w4cx/Uz6N7jklSfbty4TvvmDahjiKKkXsc+lYrXLm5sO3DUbrXEtWtXaN+8EQ2ioyiu1/O3SsW7HO5s2XeEzJmdPm4gAS5d8qNTy6Y0jo6mSKSe3SoV4TlzsXnvYavohEajkb7fd0Bz8gQdtBEEiCWsEYuZOGsu39mp9w8PD6esOg9ZoqLojkXvvhKoWLsua9Zutqp75sw/dGvfku+MRgpERrJNpUKQrwDrd+232oSPioqiW7sWvLrkR7uICF5IpawVipi1eLlV5M6vldT6He3du5shQ7rTvr2RnDkNrF+vIkeOcixfvs3ueFm7d29nxIjetG8fTfbsUaxbpyJv3m/x9d2UZBDFz6W6KYrlPMVdLGd/wBLXahqwGMsgRwKcAQanQIHjSRo6+idPHjNuzEge375JnsJFmfD71ASDS5nNZs6dO8PuLRsRCIU0btEGH58KKW6PvQQFBbFh3SpePrxL3sIlaN22PZkyZU6wnadOnWDv9i1IpFKatW5HmTLxU7WBJf7GjwN6Ex0SgsotBytWrYsXbTGtsMT06MvJvX8jlkro1KsvP/74U4J1374NYvOm9Tx7+ICS5bxp3LhZoikak0tgYCCbN67j5ZNHlCzvQ+PGzexW8bwnICCATRvXEhLwisKlytGwYZMEVTxRUVHMnj2Dgzu2onB0ZMDQkdSuXTcBi8lHp9PRrl0L/r14AYFESv+hP9G//+AE675585qN69cS8OoF5StVoV69hglGdjUYDEyfPonje3aTxdWF/kNGUqVKtQRtPn78iA1r/yLw5Ut8qtekSZPmdiuY3vPo0UM2rvmLwNev+LZmbRo1avrJNt+TmgPAp0+fsHnzOkJCAqlSpQ41atT+5Ai0I0YMZdOmFQiFRlxcvDh48MRH815/FkefhniSRo7+1KkTdG7ZhPpmMzWBI8A+gYCVm3ZQ1eZE46+jhrN3/Wp66nSYgcUKBc27dGP0r2mvUtFo7tCyYS3qGQz46PUcVSg5r1Kx89AJPDxyxtYzm82MHNSP07u2000bgUEoZJFMRpf+gxk83DoF26JF85g8djQtsRzU+huLWmPJui12x9hPLkajkRJ5c+Cg19MfS/KJP4FcBdUc+cfvI1d/+XwsYmq3di14c/ECnSMi8BeJWCKR8suUGbRt38mu+0VFRdG1dTPeXb1Mp4gIXovFLBFLmDBzDi2TiISaFAaDgU4tGhFx8wYdIyJ4KRazVCxh8p8L4iX+PnLkIP1+6ESX6GjyRUWxSalCn9eLjX8fRKVKaEfi4xw4sI/BPbvSNToaz6goNqpUmPIXZP2u/SgUn55I/Uue6VepUpbnz+/Srx+4u4OvLzx7BufOPUzybEKGo0/kgZZXe9Lj3VvipoaYDPhmceaC5nFs2Y0b1+nUqDY3dTqyxJQFAUXkcrYcPEGhQmmbGahtozo08zvPgDjPYIxIxP0Gja3imJ8/f46BbZpxTauNjbvyGigmk7P/nwtWMxW1mxOTzCbiBpIdBqxWKrn5OG1jlAwY0JvzG9dxG8tGIMAzLIGpVm3eQdWqX07iFXtI6ju3ffsWlv3Yn3/irH3fASrI5Vy6dd9qmSe5bNy4jg0jhnBcq43dTLsFVFYouHTrAQ4OKRcyrF27ip0//8RhrTZ2k/kqUFOp4sq/D2OdbXR0NOWL5mfNu7e8jxRvAprLFZT+aRR9E5lVJIXBYKBskXxsCw3h/ZzZBDRSKKj08zh69OybYpu2fKmO/tatm9Sv/y0XL0LRmHNs0dHg7Q1hYQU5fTpxxVNGmOJEePbuLbZfmb4x5XE5fPgAbQyGWCcPloTWrY3GNE/kbTAYOHXxAj1sXrR9jEYOHjloVXZo/x466nRWwbWyA00EcOTIh8QRz58/I8xsit3gfE9/INw2h1kacHLfHnrzwcmDJVF0LeCPP9I+/MR/yeEd2+gex8kDFALKSiScOXPaLptHdmylRxwnD5bIjMXFYi5cOGuXzUPbNtMzjpMHKAUUFAm5ePFDlM3bt2/iEGWIdfJgcRZ99DoOb99i171v3LiGi8lI3IVRIZYwyYe32Wfza2HMmJEUK/bByQOIxTBwIAQF3Uv8wlQiXTp6MWAbpzKU+BIjuVxBaAIbK6EiEUrlp08jk0IkEiERibDd0AgFlFLrNWWFUkmoKIF2CkVW012ZTI4AsM0DFMrnedBCiYTgBMqDwa7R59eEXKUiJIHTjaFg976DXKUiJIHyUDMoFPadclaoVPF+G2Yg1GS2+i4pFErCjSZsc3mFAnI77y2XKwgzmbCdv4cC8s98avtzkzlzZkISeJghIWA2p/2vM106enX+AoyE2MiMRmBETHlcmjZtznaBgLjBea4Au4FGjZqlaRtFIhHNGjZhjEQa+2OKBn6RyWhhE/yseYtWrBWL0MQpOwccNZutlBIuLi44yWT8zIdoj1FY+u6cQGyW1KZ7/8EsAKtk2scBP2DSpLRPrvBf0rJTV+bKFcQNjLsLeCGWJBr35GO06PQ9sxVK4gY53gK8lcspX94nscuSbmeXH5ipVBI3ueY6QO/oYJVyMX/+AmTz8GBhnJdXCDBZqaTF9/YlnClSpCgqVzeWxrEZDExVqmhpp82vhdmz5/PiBWyJM3F58wYmT4bSpeMnU0lt0qWjX7Ptb46pVOTCkvQjF3BMpWLNNusAvu7uHkyZvYAqcjmNHRxo5OBATYWCGfN9P0vEwfHT/uBaoUIUVqno7OBAPoWSsLLlGDrKOsCUl1d+xk6ZiY9cTjMHB+o7ONBAqWT+sr9wcspiVXf1zn2sFQjwBFoB7sAFsZg9h06keX/69RtIvmLFKYIl8UoloD7QuvP35MmTvqNmVKhQkc4Df6SoTEZrlYrqjo70ypyZZeu3JJrT+GNUrVqd1r37UUgmo43KgaoOjgzKkoVl67fYndO4Ro3aNO7WE7VMRhuViiqZMjHSOStL1221UpQIBAIWrt7ATFc3vB0caa9UkV8mp3zLNnz3XSu77i0QCFi8ZhO/Z3PhWwdH2qksNqu060CjRk3tsvm14OTkRKtWP9C5s0Wb36QJeHmByeTA+vVpv2yVLjdjAUwmExs3ruXChfOUL+9NmzYdEpRGRUZGsmjBXDatWo5AIKBt1+707N0v1eReH8NsNuPnd4GAgOfkypUvSRnk27dBHDtmSYZeo0btRJdDLlw4x6C+PXnt/5oC+Qow33eFVWzstOb8+XNMm/Y7zs5OjBs3KdFoi8uXL2HxtElEhIXi6pGTcdNnx1NFfUkkZ6Pv9etXnDx5HEfHTNSoUStRaeeJE8dYMXcWb168oEylyvQZPMxKaRWXly9f8M8/J8mc2Ynq1Wsm+t08evQQK+fNIeDVK8pVqUqfwcPIkcM9wbovXjznn39O4unpwTffVEjU5v79e5g96TfeBgVRsXp1Rv4yATe37PHqnT17mn7dOqMNDEAoFFK+XkNWrkw4/G50dDQnThwlMDAQH59vE82p++bNGxYt+oNz5w6TNasLnToNom7d+gnW3bNnN2vXziUsLBhv71r06vWjVTjjlGI0GunSpR2XLx/EZDLh4ODGokVrPym3gNlsZsWKpcycORaDwUCFCtWZM8f3o+dGMlQ3n7i7bjab6dC8IcIrlxgSI6+coVAgLe/Dyk07Pjm8bEpILbXAkSMHGfhDJ37R6ylnNnNYKGSOQsHGXQcoXjzts83HJak+jRs3mvUL5zEZKAnsAWYBSz+DDNReUusZrVu9kpljRjJBp6UgsEMsZo3KgT3HTif6UvwYK5f5Mn/CWH7VaSkAbBGL2eSYib3HzyTq7CHpPi1dNJ8lk3/jV50WL2CTWMK2zJnYd/yc1Yz38uWLtKhXg/ZAZyxKqxGAInce/rmY3Kjl1gQEBFC/vg+NGwfTtm0UT57A+PFKOnUaQZ8+1mcy//xzKps2zWL8eC25csHatVIOHHBm376zODtntev+FSoUwmh8yZQpFink8uWweTPs2XOaIkWKf9xAAkyZ8gsHDixh3Dgtbm6wcqWUM2fc2Lv3bJKqrAxH/4k/upMnjzOuSzuuRkTEbtRGAcVUKqav20KFChWTujxVSa0XV41yJZn+9DFxxz3zgH1Vq7Ny885Psp9SEutTdHQ0hTyysttsJm6U/CnAMrfsnLlx97O1MSWkxjMyGAyULpyXw2FhxHUXI8Rigtp2YPIfKc/Hqtfr+aZwXk5FRBBXEPyjWEJkl+/5dXLi+yOJ9Umr1VK6iBfntVqrVI79JVJE3Xsy5tdJsWU+pYvy7fNnrI5T7zmWeCn7j9vnGKdMGU94+DwWL/4Q+fPxYyhdWsGlS/dxcLBo0EJCgilXriA3bujJGWdC1LWrjBw5hjBkyChSyuXLF2nWrAaPHkH2OJOXVq3g2jUvTp++mmKb/v7+VKpUlLt3I4k70WjVSkHJkr/Qq1f/RK/NkFd+IhcunKOJjXxNAjTW67lwIe0zyKc2ERHhPHz5HNuzmC2A8xe/nATVGs0dTGYzlW3KWwD+/m/+iyZ9Nh4/fkQmkxlb19ciOhq/U/bto9y/fw83gRDbUx8toqPwi5NIOyVoNP+SWySOl6+3RZSBC8ePWpWFvXxBG5t6OYGCwKpVK+26/8WLR2nRwjq8s6cn5Msn4fbt27FlN2/eoGhRqZWTB2jRIhI/P/vC/65cuZwSJaydPEC7dhAS8sQum1evXqJ8eRm2q0ktWui4cOGwXTZTwv+1o3d1deNeAqfx7snlCa5DfunI5QpkYgnPbcrvAW52TmHTguzZsxMFBNiU3wMUdm5cfi04O2clMDoqnqz2HuCWxBJLUri4uPAqyoDtSYl7gKu7h502XXluMKC3Kb8LuNnYFEpl2M7BorEs4RQtat8yh4uLB3dtjBoM8PSpwSpmk6urGw8fRmGbp/7uXQFubgnveXyMIkWK8OgR2Ibnv3MHhEL7pLJubtm5f9+IyUaveveuCFdX+5brUsL/taNv1uw7TorEbMAiRzRhibV8XiT+KlUAYrGYTp270luu4F1M2XNgsFJJVztOMqYVWbNmo2DefHSDWJ34Y2AAUK9dx0SvSw9ky5aNGtVrMkAmiz3voAF+USjpOsC+Z+Tmlp1KFSszSCqNdfa3gfEKJV36DbLLZs6cuShbrjxDJNLYAFc3gImK+N+ldr37MQG4FPO3HstpbLNQSKdOXe26f5cuA5k8Wcm1azE29TBsmJTixb/B0zNvbL0CBQqSP38RRowQo495K126BDNmyOnUqZ9d9+7duz8Gg4BRoyAy0lJ24QJMnQrff5/cmI3WlChRiixZcjNunIj3eWjOnoV586R07NjLLpsp4f/a0WfKlJk1W3cxIXceciuU5FYomeqZl3Xb//5qD/iMGPsb2Zt+h5dMRjFHR0rIFdTq0YcuXbv9102zYsv+I9z3yEkOLJs1hQGvylWZOvWP/7Zhn4EZC5YQXLU6uWQyijg4UlGloteY8dSuXc9um7N8V+BfuSo5ZTIKOzhQVeXAwF8nUr26/RrtP5ev5sm3FWNt1nRwZPikaVSsaL3oNnr0WEpVqkoVLEs22YDVIhGrd9qXjQnAx6cCo0bNpG7dTBQq5ICHh4x79yowf378BDqLF2/mxg1vcuaUU6iQI40aZWb8+LmJBv1LDsuX72LpUiHZskHOnFC9Onh710k0QN/HEAgErFixg1OnSuPhIadAAQdatcrCtGlLKFq0mN3tTC72xdtMR5QqVZrjftd58OA+AoFFs/451TapjVgsRmcyQlQU/0ZG4iyWECUQfLY+PXnymImjhnHwxDHkEgnffdeK0b/+Hk9V4OSUha59B7Jg5lSev3tLca98dO83KEEJ7MOHD5g4chiH/zmBQiqlVau2jBw38Yt7Gd+9q2HiyGEcO3saB7mctu068NMvE+IF63JwcMR3zSb8/f0JCPAnX778iZ6e3bNnN2MH9eFlaChSLCPDtdv+jhfxUKVyoERZby5e9ONBWCjlihankJ3qkLjtLFnOm8tXLvMgPAzv4iUpmEj8p83bdhMSEsL69avJn79gqiin2rTpQPPmrbh//x7Ozs4JJmQHyyxp7dp9vHr1EoHAQNasHomeXVizZiXTp4/kzRstSqWAsmW/Zc2anfHkpZUrV0WjCebgwf08fHifdu06kTlz/KiyKcHJKQslSpTn1q3bPH2qpWpVH/LlK/DxC1OB/2vVzZdEavWnV7fOXNm9gzVAWeAw0AnoOWI0Q4eO/GT7SREaGkI171L0eveOfiYTYcAvUikPixZn6/6jVi+bebNmsHPODHy1WkoA+4FeCgW+67danSR99+4t1X2+YUBICL1NJkKA0TIZr0uVZsPutI1HZEtSz+jNmzfUqliGEWFhdDObCQB+kskxfFuR5Ru323W/K1cu0bxudaYAXYE3WJa37rm4cvbWfau6v44axrV1q5mv01EA2AoMVijYsvdIkiPGpPr089CB3Nuykbk6nUVeCQxTKtlx4DhqdSG7+pTWJNWf/fv30rt3WxYssChoHj6EHj0gMrIgBw6kfRrFnj3bYDIdY+ZMPW5usGIF/P67I4cO+aVIApuhuvk/Jzo6mkN/72Q7UBFLSsGGwFLgr9lpvySyacM6Kup0jDGZyIIlt+kKg4HXmjv4+X1Q/URFRbHgzz/YpNVSHpADzYDpOh1zJ02wsrlh3Rpq6PWMNJlwwpKVflVkJA9vXOPq1ctp3qfksnrFEppHRvKj2UwmIB+wPlLPxXNnuHtX87HLE2T8yGF0weLcHYH8wHYgIMCfE3HUNCEhwaxds4qtOh0lsASV6wj8FBnJ4ln2BZMLCrLkCtcMHCsAACAASURBVNiq01EMS0rGrsCPkZH4zplpl83/milTRjBqFHTpYkkLWKwY7NkDd+7c5cGD+x838Ak8fHifM2eOsXGjnvz5wdHREtCsTZtI/vprcZreGzIcfbrC3/8NBrMZ27O1lYCQSFv9ROpz7+Z1KtlEyRQCFTFz794HZ/f2bRBCYzS2Z3UrAffuW0st7l6/SmWdzqpMBFQQCOx2oGnB/evXqPR+5y4GKVBeLLG7nW8ePeB/7J11WJTZ+8Y/00GIYLt2YGF39+ra3V1Ya7drd3d3Fxb2unYgdjtgIeCKjcAEU78/RmIGUBwBXX/f+7r8w8N5n3POxDPv+5z73LftOWE5FrVJH58YRcwXL17wm1SCrX9YJZMJ//v37Bo7IOAZOaUybM9sVjIa8bt3x66YPxofPryicmXrNldXyJIFrl69kqxj+/v7U6KEFNsKXaVKkfj53UzWseF/if6XQrp06ZEKBNy2ab8IOMu+z7kpMchdyIOLNsqGJuASAnLnjknrqVO7YhKJsRVnvQTktqlZ5i7kwQW5dY3bCPiYzSkq6/A15PIozEUbuYNIwFevJ29ed7tipsuekzM2bTos+vFlypSPbsuaNStBkfo4dNWLAgG5CxSwa+xs2XLwNFIXzd6KwgWhkDx2UiZ/NFKlSs/589Zt799bzD++R9ogMciTJw/Xr0dicy/AhQtS8uRJfve3/yX6XwhisZgadevTFIu6pQFL7bs70MFO6t63oGWrtpyXy5kuFPIJCAa6S6Wky5OX0qXLRPeTSqV49h9IK6WS61hOI3sDQxUK+tkIurVp15GTchmzBQLCsHCzu8hkZC1QiKJFiyf7mhKLjl174CWVskggIAJL4bS9TE6x0mXtTvTjp89mA7AMUANPgeZY6KlVq8aYuKRK5ULrNu1orlBwH8uPwXZgplxBLztZIm5ubjRr3ormCgUPP8fcAsyTyek5YIhdMX80RoyYzrRpsGWLhTb58KFFXCxv3tzkzp28m6I5c+amXLnKtGkj5+lTiIiApUth+3YpHTv2TNax4RdP9GFhn3j48AFhYbYK3MkPk8nEmTOnuHjxPCbbUxI2CA39yL179wgP/7Ldrtls5skTfwICnifYZ+W6zRRv0ozfhUKkQHuxmHb9BjJ8+Oh4+0fFfPHi6yf+AgKec/iwN+/evY3376lSubDv+GkuVa5GWpGIAnI5NG3JRq+DcVg//QcNo8Wov2iWJi1ygYAJufOyYN1mKla0frZ2dXVj37FTnKlcFTehiMJyBdIWrVm/a993M4kMBgMnT57g+vXvtzlMnz4Dew6f5FCZcjgDhWUy0rTrwPJN2xO8xs9PxdGjh/n0Kf7PZ4kSpVi0ZhNTlQ44YKGghhT04PjluHsT46fPoULfAdRwdkYpELCoQCHW7dxLoUL2331Pnr2Akp79qO6cCqVAwIrCRdi050CyO6/Zi5CQVzx48ACD7empz6hXrwETJy5g2DAFCoVFRVIkKou396UEY7569S8q1aMEY0YhODiIKVMmcPTo4QT7LF68hUyZOlGqlAMuLkL27y/H7t1/Jyhml5T4JVk3JpOJqeNGs3nTOjKIJbwy6OnQoTNjJs/4bnPfxGDHjq2MG9QPkdGIERBJJMxbs8lKOx4siWbCyCHs3LmNTFIpr/QGuvXszbAx4+MksWvXfBncswvh796hN5vJnDUri9dvjVO+0Ov1/DVsIF5eu0gvEvPGbKJX7/4MGjEmTkwfn8sM9eyK5uMHdCYTWbPnZMmGLeTMaX3wPTw8nIbVK/D4+TPSYjnRWrFCZbZ4HUzw9TSbzaRL55woJpHZbE5U0k5sv8Rg/vzZLJkxBaXZTATgqFSy3sv7i9zrrzGjZs2axoq5M3E0mwkHnBwc2LjvCEWLFrPqFxISQpMaFQl+HYIbFvvKeo2asmz1hjgxp06dyNpF83A0mwkDXBwd2eJ9Ig6TJizsE4M9u3H23BnSisWECoWMGj+Fdh27fPF1SCzbKylf+6TGu3fvGDSoE76+V3B1laBWi5k4cT6NGjVL8BqTyfTFXPDmzRsGDOjAjRvXSZ1ajFYrZfLkRfEepCxXrhAvX77A1dVSCpJKBezadYpixUokGP9bXs//sW4SwNKFc7m+eQMPtVoehofxUKvlxpZNLEkBtsCzZ08Z8WdvVhqNvMHyJZ6u19Onc1s+2FgZzpk+ice7d+Cv0/EgLIw7Wg2nV69g3RrrXfi3b9/SsUVjpgQF8kKjJlirobu/H60b1UFnU/SbOWkcQXv38ESnQ6WO4JZGw9HlS9i8ab1Vv5CQELq0bsqsl8EEqNW81GrpqHpIqwZ10Ov1Vn1b/FGDjM+fEQy8AB4Czy+eY8ighIWYviUpJLZvUiWas2dPs2j6ZPaZzfwLvAf6qNW0afD7V5++EsLJkydYMWcGhz7HfAd0i4igdb2acWI2qVmJUq9DeI2lFHUNOH9gL7NnT7fqd+jQAdYunMvRWDHbh4fTsk61ODEH9+qK69nTBOl0+EVEcDIsjHl/jbJi53wPftYkD+Dp2ZJ8+S4THKzD3z+cffs+MnZsH27cSJgy+bUbvh49mlG06FWCg3U8fhzB7t0fGDmyJ3fvWu+AtWrVhNDQF1y5AkFB8Po1NGxoplWrL/sjp/Tr+Usm+vUrlrJcoyZKrSYDsEyjZsOKpck+9tSpEy1mG4AAywvcAyhpNjNjxtTofiaTiQ1rV7NKoyFK5+g3YJFGzfolC6xi7tm9gz+MBpp9jikCepvN5NFqOXHiaHQ/g8HApk3rWKXVEKVskxVYqFGzfvF8q5i7dm6lidFAo1gx+5vNZNForHxow8PDuffoIeshmoGRHUvd+MSenXa+Sj8Ws6eMxxOogWXtEmAskNpgYP36NXbFnDN5HP0h2mNVCkwClHo927dvie73+LE/QSGvWAlEHfcqAMwBdqxYYhVzwbRJDMHCRgILXXYaINLp2Ldvd3S/kJBXnDt/liWRkTh8bvMAJsTzvv9qePzYHz+/e8yerY9mtJQuDUOHatm0yb7v+4MH9wkM9GPGjJiY5crBgAE6Nm1aZtX31q1/mDULPD5XyBwdYflyiIw0c/DgfnuXleT4JRP9q9CPcah77sC/H205BEmPkJfBxKf6XghLHS8Ker2eMK2GHDb93IFX79/FiZlPG5ce6a7X8+pVjHmdVqtBp9eTNb6YNnX1kKBA8tlSAAB3o8Eq5ps3rxFgcaqyjam2ufP/ryA0JARbLooAi5n306f28anD3ryJox4pwKLgGJuj/fTpE1zByugdLK+n1oZGGv7ubZyYQix8+idPnkS3vXnzhowSKbauq+7A65fB37aQ/xhCQl6RM6cEW+vnfPnMhIQE2h0zd24xtiZe+fKZCAl5Eae/u81eu4MDpE3LT6WA+0sm+lIFCmL7W7oXKJ0CmhKVqlZnOzF+tWCh2e0DatWKERCWyWTkz5Yd262bfUCpItY13ZJly7PPwcHKqFkHHBaKKFUqhs3i4OBI9oyZOGET0wsoVaykVVupCpXixNQAR8EqZrZs2ZELhXFofl5Aup9IEfNbULB8BbaDlUl1KHAOaNDAPq9g99JlsX2+eY+F/dSwYZPotrJly/MRsGW37wYyZLW2W8xdohQ7bPq9xSIeVq9ew+i2XLlyE2I2YcvW3ysWU8Jmc/tXQ8GChbh3T0ewze/Z3r0ySpSwz62scOGi3LihI8RGMXvvXjklS9qWZKRWPrBgUbl8/Rp69uxt1/jJAdGECRN+1NguwECNJhJ79oMdHGSo1ZHx/i2He34893shMRqRADsFAkYrFMxevoYsWWzvd5MWpUqVYfHKpZzU68mMRda1GxDh6sbiVeutanO/5cqN5+GDKIxGhMAWoZCJcgXzVq2zOhKdM2cuvI4e5lDoR9IbDKgAT4WSXJWr0qNP/+h+AoGA33LmxPPoIRwNBgTARqGQaQoFC1att5JezpUrN9u8D3A87BPpDQYeAL0USgrWqE3nHp5WMT+q1Yz29SHKnXYdlrLE9MUrv3gU/kvv0Y9EmTLlmbpqGXdMJtJjMYRvB6TNk5dR4ycneN2X1lOmbHkmrlrOfbOZ9MANoC2QOX9BhowaG91PJpPx5MljJj58QAYsP9gLgaXA8o3brRymypQtx/jVK/Azm0mHxWS9DZCzSDH+HDIiup9EIkHp4MjAyxdJY9CjARaKxWx3dGLBirVfdC/6Wd+jxEIuV2AyGRg16gbp0un59AlmzpRw4kRq5sxZgVJp+5zzdSiVSrRaNWPH3iZ9ej0fP8LUqVLOnXNl9uxlyGOd63B1zcCsWUd4/x5cXODMGWjbFlKlysiwYWMTHuQbYPseCQQClEopWD46HxMT46usG3d3dzcs6r25sNyc+gO9VCrVG3d397LASiynrp8D7VUq1euEYtkgO3awbsLDwzh9+hTOznKKFSuLs3P8QkN3795mxrhRPFWpyJHXnVGTp+PhUSTR43wPPnx4j2ePTty4dBEBULZ6LVasWh/vh+7GjWusmDuLgCd+uBcqTO8hI8mfP+4hF7VazZKFczm0YysikYiW3T3p0bM3YttnViyPjDMnjCHg+XMKehRh9KRp8SbkiIgI1qxcynGvXUgkUpp26Ub79p3jNZ5et241K2dOJTzsE24ZMjFu1ryvCld9jdFhNpu5cuUygYEvKFy46Bd/NN6+fcv582dQKJRUrVo9QRGwxCIwMIA/Pbvjf+sGYpGYas1aMHfuong36UwmE5cvXyQi4gO5cuVPUIjq2bOnDOzbgye3biGWiKnVsg0zZ86LN+b8+bPZunQRGnUEGbPnYNqCpZQuXTZOvydPHtOrc1uC/P0Qi0T83ro9s2fPjzfmyZPH2bBoPiGv/qV05ar0GTTsq9S9pNSMCg4OwsfnEqlTp6Zy5WrxfjaTC97eB9iyZRGfPn2gbNla9O49xEq3/lthNps5eHAfW7cu4ePH91Su/AeenoNJkyZNnL5bt25iwoSBCAQWCmaBAhXYv/9onH72IkWsBN3d3V2BwiqV6szn/8/Gsi/XA8sNa2eVSnXB3d19LJBTpVJ1TeT8s/ONif7IYW8G9elBaZEIkQAuGwzMWbyCBrEejcGygdi9TTOe3r1NWYEAH7OZHIUKs3bH3hRRPNzrtYuRg/pRXizBAFw1GVm8aj21a8dvbAxf/8Lt2rGVscMHUUEsJhIB100mlq3bTPXqNa36ffjwno7NGvD+6VNKCOCcyUTxsuVZvmlHgkbVyYUvren169d0bFoPTXAQhYGzRiMVq9di4eoNcZQH165azszJ46gikRCKgAcCAWu27qZs2XLJvoaXL4Np1/gPhG/eUFAApwwGav3RgDlLV8X7g5jUaN34D3wuXaAq8C/gJxAwd/kamjZtkSTxk8rCcsaMv9i4cSXVq0sIDBTw6pWCLVsOpbj4WUqLHZpMJsaOHcD+/TupXl2Mvz9ERLiwdeuRBE3PvxU/xDPW3d29GdAbGAWsV6lUhT63pwGeq1SqxGbS7HxDon/9+jWVShbib62WKHbqbaCaXM6ZK7esSh1jhg5AvXMbG3U6RFjq5Z1kMpSt2jJ1zsJETs8+BAa+oFaFkpzVaonaEfAF6igUXL75IEGz4i99QJ89e8ofVcpyXquN3py7BNRXKPG984hUqVyi+/br2oE0x4+wRK9HgOURrJlcTj7PfgwfPS6JVpk4fGlNXVo2psCFc8z8XGLSAvUVCioOHUnf/jHmDnfu3KJ9g9+5rNGQ/XPbcaCjkzPX7j/+7jv7r6FVvVpUuXGNcUYjAiAC+F2hpOH4yXTp2iNZx162bBErJozlJkQzyHYB3QQCHgW+iSOtaw+SIjEeP36UKVO6cOGCmqgb3rVrYc6crJw9ezdFqYQpneh37tzG+vWDOX1aTZRy9Jw5Qry8CnLw4MUvX5xIpDiP3t3dXYglyR/EwtyLPk6pUqneAsLPTwBJDm/v/dRHQOwjCEWApmYzBw7steq7e/dOpn1O8mChDk7T6di9O/npgPv3e9HSZCL2tm9poLZAwOHD3nbF3Oe1m3YGoxUDozxQTSiwOomn0+k4dPwIkz8nebDQ/KZotezZstGusZMDYWGfOHvxPOM+J3mwiHVN1GjYs3GtVV+v7VvopdNFJ3mA34ECZjNnbLxLkxqvX7/m5u2bjPyc5AEcgHEaNXvWr07WsQG2r1rOSGKSPFhouxnNZjZv3pDs4ycWe/asYfjwmCQP0LUrGAzv4vDOfzXs2bOasWNjkjzAwIEmnj798gn2lMa3FtEWA+HAEqDJV/omCp9/mb4KodCIizHuMeTUBgMCgZG0aWMIaxp9JLaVexdAHamz6pccEGAgdTy0w9RGI0Kh8YvjJ/g3k47U8a3dZLKKGREhxGQ2Y/uKugCaFFh7fIh/TC0igQBbt14XQKuznqdJryV1PIeYUgtALDYl65rCw9+iEImwvW92AXQ6bbK/niZ9ZPQGuO34BoMmycb/3jh6vRpbTw6BAFKnFiGVfn/8b0VKjhcZqcHFxbpNLAZnZzFyueCneY8Snejd3d3nAHmABiqVyuTu7v4Cizx41N/TACaVSvU+oRjxIbGlm7Jlq9BSJGKCXh99cOcjsE0qZXP5qlaPNjUrVGbFudOMiFWWWiEQUKtilWR/rCtXviq95HMZpVFHc6XfAnsFAvaXqZzg+F965CxfuSYDly1jqFodncRfA/tNZnqVrmR1XcmChdhw+xaxZZKWi0TUqFE7xY1aEl6TnJxZs7HrsT9tYrWuEEuoVruu1TUVav7BnD1eeKojiNphCABOR0YypUiZZF2Tg4Mbqdzc8A4KIorMaAZWSKVU+6Nhsr+epWvXZfHWTbQl5ov6AIt367rGrZNk/KQodVSu3IRVq27QpImGqD3i69fh2TMTOXLkT9HPXUqXbqpWbcyKFU+oWlVHVIXqn39Ar1eQJs1vyfIexSrdJBqJKt24u7tPA0oAjVUqVdQpm+uAwt3dPergnicWOnCyIH/+ArTs2JWSSiUzBQJmCwSUVCpp2KptHDbNXzPnsjBVKjrKFawGOsoVLEiVir9mzkmu6UWjePGS1GzSjJJKB+YA0z/Ps333XnYr5JUpU46K9RpRSunAXGCaQEBJhZLuffpbGSUDTJq/lLGOTvSUy1kNtFQo2eHqxvAv0AZ/BKYtXsGfDg70kclYBTRWKjmWLh2DRlqrV9auXYccFStRxsGBRcBEoZCyCgXDx0wgbdq08cZOKggEAmYsWUVXpZI/pVJWAvWUSq5kzEyfAYOTdWyAqVNn8dLJiRJY6JejgbJAi7YdSZ8+fbKPn1i0bdsRtTo/Vao4sGwZjBwppk4dBTNnLk1xAkBKo0ePvvj7Z6d2bSUrVsDgwRJatVIyc+bKFNmsTywSw7opiOV8hx9EG8I/U6lUTdzd3ctjoVfKiaFXhsQbKC6y842sG7PZzOXLFzm4ewdymZga9ZpQsWLleDd73r9/x45tm/G/c5vcHoVp065jghuhSQ2z2cykSX9xYPtWBEIhbbv1ZEgs3nN8SAwV8fTpfzi2zwuxVELjVu2spH9jIyQkhB1bN/Jc9YiCJUrRsnXbBGmoSQ0LZdKHv48ewtUtFbXrNk5QN/7ff1+ybdMGgp74U7hceVq0aI2jY9xHVJPJxIkTxzjpvR+lkxPN2rSniM2hsm+FyWRi5cplHNyzE4WjIwMGD6dKlfgP2AQFBbJt03o+vAqiQMlyNGveKl6qrNls5vz5s5z++ziOzs40a9E6zg/xt0Kr1TJ9+iTOHvZG4exMvyEjqVevwXfFtCir/sO5c/+QOXNG6tZtasXft3eeM2ZM4dy5Izg5uTFs2F9xlEhTAl/7HhmNRk6cOIaPz1nSps1Eixatrc6X2IPw8HCmTp3A1aunSJ06IyNGTKBkSfuNyW3xQ1g3SYjs/KKesY1rV8Xv1g16YtFaXwWUrVqDzbsS9g79mdeTWJjNZkYO6suZ/XvpqNEQLhKxUSxm1OTpdOjU7UdPLxoGg4GqJT1QvwymBxY1zvVAi45dmPUFVtbXFFP7dm3P/TOnaauO4I1EyjaRkOkLl9G4SfNkWYc9MBgM9OjRgoAAH9q0ieDlSynbt4tYuHADv/+eMP33S9Dr9XTp0oRXr67TunUEL15I2bVLxNKlm6lR4/tNwr8FX3qPtFot7dvXJyLiPs2bR/DkiYy9e0WsWbPL7h8ljUZD27Z10etVNG0agZ+fnAMHRKxfvzfJ6L9JkehT7kTD/xPs2+fFw1s38IdoYbGBQN4z/3Dp0gUr4+tfDRcvnuf8/r3cVn/enzAY6G0wUHLsKOrWaxTvYZMfgblzZ2B8GYwKovVhegMlNq1n4ODhZMqU+ZtjHjlyiKdnTnNTHYEcQB9Jdz1UHdiPmrXqpMj5jcTAy2sXHz5c5vp1NRZ2ZiQdOkCDBt25ceOpXaWWXbu2o9Ve49o1NZYjEJG0bQstW3bj+vWncc5F/Chs2rQWpfIup05pPuvY6GjeHHr16sKVK/52SZivW7cCV9cH7N+v/bw/oaVxYxg8uDMXLz76aVQ/f0mtmx+JTRvW0I2YJA8WQbAWwOrVK37MpFIIxw7uo6tGYyXYlQuoIRZx6tTfCV2W4jiycwcDwUoErABQBvvfo+N7d9ErKsl/RmGgmFjEpUvnE7osxXH8+E769IlK8haUKQM5clg8D+zBsWPb6ds3KslbUKkSZMxo4saN698546TD8eM76d9fYyVWVrs2SKVqHjy4b1fMY8d28uefWmL/RtSvDwbDp5/K0/h/iT6JIZZIUMfTroaf5s4muSCWSNHGcwejEQh+qrWLxCI08bSrAbncvs1DsVQWb0wNIJF8/8GmpIJYLCUeIVS0WjNisX3vkUQSN6bZDBqNGan053nfxWJJgmu39/MpkcSNaTaDTmf6qT7z/0v0SYx+/QexCYu/ZxTuYfFEHThw2I+ZVAqhScvWrJHJeBmr7Qpw2Wj6qi5OSqJtz97MBSsz7TNY3qdevRI2U/kSGrdpzxKlktgC0yeBpwIhFSpUsneqSY4mTTozb56S0NCYtgMH4ONHhd0biI0bd2buXCVhsUrje/ZAZKTjd2+aJyWaNOnKzJlK1LHuxDZsAGfn9Hb7+jZu3IUZM5TEVphevVpAhgxZyJEj5/dNOAnxvxp9EqNKlWrUbdyMQvu9qIdlM/Y40L5rDwoUiCtW9iuhSJFidB84DI95s6gvFBAuEnHaaGTJmg1fVFBMaXTv7snRvXvIec2XhljOOZwDRk6Ygovt6ZdEonLlqjTs0p18a1bRQCDgjVjEZbOZtZu2J4lUQVKhTp0/uHSpDXnzbqVBA3j5UsL16wI2btxhNx2wQYPGXLp0kjx59lC/vpmgIDG3bonYsmVnilh3JhYtW7bh8uWT5MlzhHr1zDx9KuHRIwnbtm23u5berl0nfH1PkyfPCf74w8zjx2L8/eXs2LH1p6nPwy/Munn82J/VS+bz+O5dcnt40KPfoGR3eo+CXq9n7twZ7Nu4DoFQSDvPfvTp82e8X6RHjx6yZvF8Ap/4kbtQEbr3HfDddwL3799j7ZIFBPipKFCiJN37DkgygaWvQavVsmTRPPbv2o5SqaBL7/60bt3+uz70arWaLZvXc3LfHhQOjjTv3J369RvGG/PmzeusX7qQ4GfPKFKuAj36/mmlgxQbFy+eZ9Om9Tg7OzNo0LCvbsImhhn19OkTzp49jZOTE3Xq/BEvXRTA1/cKG5cvIiQwkBKVq9LNs993qS1+K/z9/Th//izZsmWiXLlqdsn5xsbHjx+YPXsqly4dwcnJlb59xyTI4jl//iybNy/l/ftXlCtXh65dPUmdOmmUUxLzHt2/fw8fn0ukS5eOWrXqJIle0t27d/D19SFduvTUrl0nwU3tv/8+xo4dK/n06QOVKzekS5ceCX5GovA/emUCb+jNm9dp26QefXU6KhmNXBCLWSKVsm3f4S8a9iYFTCYTXVo1IcL3CgM0agzAXKWSnDV/Z9Eaa70ZH59LdGnVhIGRkZQ1GjklFrNaJmPP4ZMUKFDQrvEvXDhHj3YtGKLTUcpk4m+xmPVyOXuPnkp2JUG9Xk+rBrVxfPiA3hoNYcAspQPlWrRi8uwFX70+PkRGRtK8bg1c/f3w1GoIBWYoHajeviN/TZlp1ffokUMM692NEVotHmYzhyRSdiuVeP9znqw2ph72IKkosHu9djFhcH9GabXkM5vZJ5Vy2MmZw6cvkiFDxu+O/y1IijWFhn6kQYOKlCgRQvv2OgIDYepUJT17jqVHD+tS2IYNq1i8eByjR6vJnh22b5dx6VIaDh++mCTJ/memKS9cOIMdOxYwZoya9Olh/Xo5fn6/ceDAeRwcHBK87n+JPoE3tHmdanS+cZ0usdrWAxuKl2DPsaQxS04IZ8+eZnynttxSRxC1FaMB3BVK1h08alWz/KNSGYapHtIy1vWLgGOVq7Bxj30CaLXKFmPC0yfE9kmaKxBwrnpN1mz3sitmYnHgwF7WDezLhYiI6M2fUCC3TM6hs5fJmTPXN8fcs2cn24cO5Iw6JuYHILdMxt+XrkcbyZhMJsp65GXDm9fRvq0AY0Qigpq2ZM7SlbahvxlJkUT0ej0lC+TkYGgosT2/BorF6Dp2YdKM5Dewj42kWNOiRbN58mQ227bF7Eo+fQolSyq4fv1x9B1rREQExYvn4soVNXliPVx36iQjY8ZBDBky+rvmAT9von/37h3lyuXj/n0dmT4/YJrN0KiRgvLlJ9Ktm2eC16a4euV/BZdu3aSVTVvrz+3JPvaFczSLleTB4srSyGDg0qUY2VKdTsctfxXNbK5vA1zyvWLX2OHhYfi9CKChTXsbs5lLPpftivktuHT6JC1jJXmAVMDvIhFXrtg3/qV//qa12jpmaqCGWGwV8+XLYDRhYdgee2ljNHLpXPL+uH8Lnj17isJgoKRNexuDgcunTv6QOX0vLl8+Tps21tSTnDkhXz4JuqkkdwAAIABJREFUd+7EqFfev3+PnDlFVkkeoE0bHT4+x1Niqj8M16/7UqqULDrJg0X4rU0bDZcvJ//af8lEn8bRKUY/+TMCPrcnN9zSpiUgnppfgFSCq2vMo6lEIsFRKsPWujkAcLNTrkAmkyMRiXgVT8w0dm4yfgvc0mfkeTxUwgCh0G75Cbf0GXgeD+0vQGAd09nZGY3JyCfbfoCba7IoZ9sFF5fUvNcb4lBwAwDXn+RA2bfCzS0Dz59btxmNEBhowM0tZk1p0rgRHGzAYCPE+vw5uLr+PNo9yQE3tzQEBBjj2KY+fy7EzS35y3W/ZKLv0LUHf8oVRDHIQoE/FQrad+me7GM3adKCQ0IhUb/RZmAPcEUosjJ0FgqFtGvfkX5yOeGf294DgxQKOthpKiyRSGjdqi395HIiPre9BQYrlLTv1ceumN+CVm07sEUsIuq5xYzFXzZALqNatRp2xWzdoRPrJGJ8YsVcKRDwxsGBypWrRvdzdk5Fndp1GCiVEnVv+RIYpVTSoe8Au8ZODqRLl46K5SsyTCIlygX0BfCXUkmHPj/PPL8F7dv3YeZMBY8eWf5vMMDEiSIyZ85ltS+UM2ducufOz7hx4uhk7+cH06Ypad++7w+YecqhePGSyGQZmDVLiNFoabtzBxYulNG2bfIa2MAvmugHDhvFb42akF0mo5yzM9llMjI1aMyg4d9fA/wa0qZNy9qtu+nhloYCjo64OzgwMkNGNu85EOcY/MjxU3D8/Q+yymSUS5WKnDIZBVq0xvM7EtPYKTMR16hFVpmMsk7O5JHJKNG2Pd17Jn+iz5YtO4vWbKRFKhcKOzqR28GBudmys3XvYbsphrly5WHeinU0dnamiKMTOZVKlmTPwZa9h+J4ks5YtJy35SuRRSanjJMzBeVy6vToTYsWrZNieUmGeavW8bhUabLI5ZR2cqaoXEGrAUOoX9+26PbfQNmy5RkyZDoVKigpWdKJrFkV/POPB6tX74nTd/nynZw/X5gsWRSULOlE2bJK+vefRKVKVeKJ/OtAIBCwYcMBdu92J3t2JcWLO1GjhiPjxy9MkbMG/6lEHxr6kTFDB1A4dxYyurgwclA/PnyIK38vkUio2aARObJk455WS/YsWanVsHG8J9U+fHjPqMH9KZw7C4VzZ2HM0AGEhibKWD1BVKhQCd97/izce4gVB49x8dbDeNk+MpmMmg2b8FumzNzTaMiTPSc16jWMl4b55s0bhvXrRaFcmSmaJysTRg0jPDw8Tj+5XM7S9Vs57Xubsdv2cPnWI8ZPn5NifOb06TPg7OiIX3gYryIicHRxSVBSd9eu7dQoXYS8WdPTok51fHwuxduvcOEiVKlcjSCjgUiJhFr1GkZvwsaGo6MT63bt4/ila/y1bQ9X7/oxbMz4eGmYgYEv6Ne1A/lzZKRk/pzMnDoRnU4Xp5/BYKB0CQ/ypHNGLhCQN50z/fvb98QVBReX1GzZf4TD530Zt20P1+7703/Q9x2mM5vNrF27gsqVC5A3b3o6dqyXou5O7dt35caNp0ycuJe9ey+wf/+5eBlE6dOnZ+/eM+zff5GJE/dy48YTOnfuGU9EePjwAV26NMbdPQMVKrizYsUiTPGY0BiNRpYsmUu5cnlxc3Oge/dmCcoPHDiwl1KlspI1qzN587rQtWsbDLa1pG/E3bu36dixHnnzpqdy5QKsXbuC+EguGTJkpHbtxphMMh4/jqR8+fIUL550Kpdfwn+GdWM0GqlXtTxFnz5hlD4SATBLIsEnazaOnrtilcSPHPZmdJ/uLNNoqAhcAPoqFExZttqqfGIwGKhTqTRlX7xguD4SMzBdIuVWzpwcPnM52fWk9+zewcyhA1im0VAGOA30UyhYuH6rlem3VqulVvkS/P7qXwYZDOiAyVIpz/MXZO+JMz/NwYxXr/6lQtH89DKZ6I/FimwUcMPZmRuPg6z6rl25jA3TJrFUo6YIcBQYrFCwcc9BSpWKkV/+9CmU6uWK0/79ezyNRkKBsTI5+rLl2LD7gF3z/PDhPTXKl6Drhw/0MJl4D4yWyxFXrMLqbdaWCkVyZ8Xx00dWY9HD8Qb6A6279mBGCjNkvoQZM/7i3LnVLFhgYbR4ecG4cQ4cOHA2QZlo+HlZKs+fP6NevQqMHh1OmzYQEACDBikpUqQtEyfOs+o7cmRfHj/ew5w5GrJkga1bBcya5cixYz5W8stnzpyic+fGzJwJLVpYmEG9e4NcXgRvb/v0iPz8VDRuXJVJkyJo1gz8/WHAACXVqvVi+PCJVn0HDOjM27dHmTlTQ/r0sGGDkCVLnDlx4uoX/QX+X9Er//77GPN7dsU3Ijzav9MMVHRwpMeSlVYa3TXLFGXms6fUiXX9MWBEjpycvHIruu3wYW9W9+vFBZuYpR0cGbhyLbVr2yfbmhiYzWbKF3ZnQ8grYutZegFzChbi4OmYu9tdu7ZzYMRgTkRERLeZAA8HB6Zu2fXTHLHv1asL2n1eHInVZgB+A8YsWk7r1u0sbQYDxfJl5+SnT1beumsArwqV2LQvxgd39apl3J46kZ2xzpjrgZwKBZsOn6RQIY9vnufSxfN5MnsGm7UxMXVAdrmC3X+fja4rv3gRQMWSHtwHYh9hmw9MFghQhYTyMyAs7BMlSuTm/n2tFatjyhQhKlVz5s1bk+C1P2uiHzt2IK6um5g+PeZu+907yJVLho/PI9zcLBvxISGvqFTJg2fPdFaWfkOGSIiM7M64cTFnLWrUKEqjRk+ZHMuD5/VryJYNzp+/Y9ehwkGDupE/vxdjxsQ8abx8CYUKKbh2zT/6RHhAwHPq1i1NQICW2GfTevWS4eIygKFDxyY4xv8reuX9+/eortUQ+95VAFSPCOf+/bvWfZ8/o7rN9TWAe8+fWbU9eHCParGSfHRMrcZuNbvEQqfT8eLNayrYtFcHHjz2t57nndvUiJXkwfLGVTMY4qz9R+LxrZv8YdMmBqphOYUahXfv3mKOjLRK8vB57Q+tX/cH169RQ2MtFyYBKgmFPHhwz655Prx+jRpa65gyoKJYbBVz3z4vUmGd5AFqAqYfd4MUB8+fP+e33yRWSR6gRg0TDx8mP6U4OfDw4XVq1LAuqbi5Qb58Mh7H+n74+/vh4SGL49tao4aeBw+uWbW9eRNMrVrW/dKlg6xZ4fLli9iDhw9vUqOGdTkpUybInFnM81hUpEePHlKqlBTbA8g1a+p48MA+1dBvwX8m0efMmQtfha2dNPg6OMY5iJMzQ0au2vYDctnUDHPkyMlVh7g64b4KRbILEslkMtKlSoVtFfUqkCPTb1ZtOfLk5YrNJ8QM+Eokdh1CSi5kyp2HCzZtZsAHrOweXVxSYxAKeWbT9yqQI5u1I1OOfPm5IrOmq5qAa2bsfo9y5C/IFZsj6kbgmslo9XrWrFmbUOBfm+t9+bm+OJkzZyYwMJL3NttVvr4CsmdPuGzzMyN79nz4+lq/yuHh4OenI1u2mFPO2bJl58EDHTb3Avj6ismePb9VW6pUblyxOaISGgqBgVC0aHG75pkjhztXrliXTt+/h6CgSH777bdY/XJy86Yevd76+itXJOTIYd8p+G/Bz/R5/SLq1KlHkJMz40UiwrDUfyeLRPg5OFjV3QH6DhtFN4WSKCXs60A3hZI+Q0da9atfvxF+Dg5MFokIB8KA8SIRgU7O1K1bP1nXIxAI6DNwGJ2USu4QkxA9FQr6jhxj1bdp0+ZclSuYLRSixkIXHSEWo06TlmrVasYN/oMwYdJ0vIElgBYLXbQvEC6V0r17r+h+MpmMbj16006pRIVl7aeBIQoFvUdYr71N+84ckkpZjqW88g7oK5GSJlduSpYsbdc823fuipdEwmogEouKpadUSpZ8BawYEAULFsJBKqMZ8PjzPI8DQ4AKPxFDxtXVjaZNm9OunYLnz8FkAm9vmDpVTq9e/03F1O7dBzF/vhwvLwsnPygI2reXU7t2HatN3ixZslK5cnU6dpQTHGzpu2sXLF0qpVu3P61iDho0mUmTLGqdJhO8eGGp1efIkZV8+fLbTiFR6NlzKFOnyvH2tsR8/hzatlXQpEkLK0mHvHndKVq0FF27yggJAb0eNm+GjRuldOzYK+EBkgj/mUQvlUrZc/QfblSqShqhEDeRiMsVKuF19B8UNnf6bdp1pPv4yTRxdUMhEtHE1Y1u4ybRtn0nq35yuRyvo//gU7EyaUQi0opE3KhUlT1HTqaI4mB3z760HD6Gui4uKMVi2qRNx6Dpc+JYzzk5ObPnyEn+LlOO1EIh6UUi/KvXZIf3iQQ3jNVqNYGBL4iMjIz371Ewm828evUvb9++/e715MqVmxWbdjDDwREnIANwMkNGDp65HIf1M2TUX1TrO4BKjo4ohEJ6ZczIxEXLrTahwUJX3eV9nB3FS+IsEpFVIuFj3Xps9Dpo9yZ0+vQZ2HHgGJuLFsNJJCKHRIq6XkPW7dwbp+/le/74Kx3wAKRASyBvqTKsW7fFrrHtgdFo5Nixw9y9eyfBPpMnLyRv3q4UK6ZAqRQxdmxOFi/eavedqj0wmUwEBwfx6dP3713kz1+A1at3M21aXhwcRBQqJCdjxvbMnr06Tt8FCzaQOnVrChSQoVSKmD07P+vW7YuzCd20aQsGD55M9+4SZDJwd4ePHwtx8GD8bK/EoFixEixevJWxY3OiVIooVkyBu3tXpkyJa0m5bNkOxOJm5M4txdFRxPLlHmzZ4p0igoP/mc1YsChSDunVhYcqi0VXnly5mbtyfYJiXWazGbVajVKpTDApqFSPGNqrK6rHfpjNZgrky8/cletTTOkyap4ODiIiIowJzvPu3TsM8+zGs4DnmDBTuKAHc1eui2M+bTQamT5hLJs2rsVBKCRSKOTPwSPo2ffPOLFv3brB8D49CAx8gcFspljhosxduS5e6uK34uPHj2TO7EZEhDHev+v1eiaPGc727VtwEAoxSiQMHT2eTl0TPjyi1WoRiURJauig1WoRi8VxOPmx/z5+xGB2796Bg0gEUiljJs+gddsOSTaHL2HgwL54e29GIACdDpycJGzZcpzixW1FFCwwGo3odLpEq1Em1WbsiRNHGTfuTyIiPqFWG6lVqyazZq1MElN6tVqNTCb7KgvOaDSSKpWM8PCv0yXfvXuLk5Nzkt7QJXaeBoMBvV4f5wY1Ify/2oxVq9W0bPA7be7d5XVkJK91Oro8fECrBr/HyycHS3nEwcEhweQZHh5Oqwa/0/nhfd5ERvJGr6ftvbu0rF8btTo+n6jkwdfm+eHDe9o0/oP+/ireROp4HRlJg9s3ad2wDnqbot/c6ZO5tWkdd7VagtRqzoWHs23OdHbt2m7V7+3bt7Rr2oChj/0J0ekIiYyk1s3rtG7w+3fzigFcXFy+mGym/jWSpzu28UinI1ij4e9Pn1gx8S+8vROmTMrl8iR37ZHL5QkmeYBxwwbyft8enhoMvNLpOBwWxuxRQ1PEGnHjxvXs37+ZXbvg40d4+xZatdLTunXC5TqRSPTdksPfirt37zBoUGdWrw7h3381BAVF4up6kr592yZJfKVSmSiqs0gkSnTydHNLk+RP7Ymdp1gsTvQ8kwr/mUR/5Ig3hbRa+pvNSLCwOTzNZkpGRuLtvd+umN7e+ykZGYmn2YwYC5ujn9lMIZ2Ow4cPJuHsvw9ee3ZR06CnMyDCUkIYZjKROTyMv/+OEUQyGo2sW7OSdRoNUdtA+YBFajVr5s2yirl75zb+MOhp9zmmDBhtNOL6KZQzZ/5J1vVotVq2bdvMeq2GDJ/bPIC5mrjz/JEIC/vEvv1erNVqSfu5rQQwXaNhzfw5yT7+/PnjGTgQ6tSxCGA5OsLChSAWm1i6dFGyj59YbNiwmMGDdVSvbpmnszMsXRrJzZtXefbs6dcD/A/Jjv9Mog8ODsJDG9eV00MdQVBQoF0xg4ICKaSOiNPuodXw8qWt3NiPQ1DAczxsaQWAh15PcHDM2tXqCLSRujh0QA8g6HWIVVtwwDMKx2Og6WEwEhQUFKc9KREa+hEJFtP02CgEBP/7Mp4rfgzevHlDarEYWzk2DyA48EWyj28whFOkiHWbSAT58sGdO7fiv+gHIDj4KYULW1MMpVLIm1f6U32P/j/jP5PoixYtzlG5nNgVXxNwxMGRYsXs23AqVqw4Rx0cif0RNQJH5fIU3cT6GoqXLsMhBwdi72TogeMisZW0gqOjE5nTpuOMzfWHgGKFClu1FS1dNk7MSOCEQGD365lYpEmTFrlSGYcCe0ggSHZjmG/Bb79lQS0UYnuiwlsopGis07vJBWfn39i3z7rt40e4cYOfSr+naNGKeHtbl0Fev4Z793Tkz/9r22f+V/CfSfSVKlXBNV8BmsnkXMZCRWwpk6HIncduimG1ajVR5M5DS5kMH+Ay0FwmxzVfgZ9KZKlu3fposmSlvUzGVSySDo3kCvKUKEWJEjFaGQKBgJFTZtBOoWAL8AhYDIxQKBg8frJVzAYNGvM+YyY6S6Vcw+KZWl+uoEi5CskusiQSiRgxYSrNFAp2AA+xmKNMVigYMGZ8so79LZBKpQwdPZ6GCiV7gAfAdKGQBQoF/VJAIG/Zso0cPAgDBliUDk+cgMqVIXVql5/KbL1r174cOuTIyJEi7t6FI0egdm0lXbr0slue+n9IWogmTJjwo8Z2AQZqNJFxNJrjg0AgoEGzljw2GJj0xB8vR0fKdOrKrIXLE/RnfPzYH5/PhhtRR6ZjQygU0qBZS56YzcwJCsTbJRUVuvVk2txFCW7U+PmpuHLFB6FQkGQf4kePHnLz5lUMBnO8dmoikYhGzVtxOyKCac+f4e3kRLVefZg8a16cjUR393wUKF6ScT6XmK/VEOiej5WbtsdhaYjFYho1b8XdyEhmBwdxwi0NtXv3Y9zUmQluKN2/f49r13yRSqWJMtF2cJChVsdP7yxYqDC5PAqz4MEDlukjiShTjoWrNyYoafDyZTDLli3izp1beHgU+eIGalKiaPESZHLPx4pnT1krFECV6sxfuY68ed2TfewMGTJSrFgZ5sw5zOrVkXh5gatrMU6duvbdOkw3blxj7NiRXL3qS5kyFb4rnoODA/XqNePUqQ/MmfOGK1ey0rnzWDw9B8RLMDCbzVy96sudO7dxcUn9RRu9xMJkMuHre4VHj+6jVDqn+Ib0vXt3WL16BYGBLyhUyCNBYoXRaOTKlcs8eHAfNze3RG3K2n6PBAIBSqUUYCGQKAXGRNEr3d3d5wDNsFAiPVQq1b3P7c+xnI2JKvaOUKlUibVLyc430itPnTpJ/+6dyCsAkUDAfaOJRavWUatWHat+Go2Gfl3a4XvpIiUlYq7pDZQqV56lG7bZvdsdERFBn05tuOl7hRISMVf1espVqsqSdZsT/KH5GsLDw/Fs35J7N69TQiLhSmQklarVZOHqDXF+aLy99zO0vyeFRSIigacCISs3bad8+YpW/fz8VDSoXAahyUQRLKc4lXI5F+/6kyqVfVS30NCPdGvdjGcP7lNYLOKKXk/tuvWZs3TVFxPul6h7Hz68p1urpgT6PaKgSMyVyEjqN2rCjIXL4iSdgf092btzG8WwHBYLEAiYuXg5LVsmDasjsfgRujCrVy9m9uzJlC4t4dUrMxERTmzYcOC7/H+rVi3G06dPKFnSossSEgKTJi2iQ4fOSTfxBBAQ8JxOnRpiNr8lWzYhPj46PD0HMHhwwlovX8PTp0/o3LkRItF7smYVcvmyjn79hvHnn8OTcObxw2Qy0bhxVW7fvkWZMvD4MajVEnbsOEnRotZPxirVIzp1aoSjYxgZMgjw9dUzbNi4OL66tkhJeuV+oDLEMW4CaK5SqYp+/pdsnljv3r2jd5d2eIWHcSEsjLOfPuEdEU6/7p14/fq1Vd9Zk8chuXiBF1oNh8LCeKHVIL10kVmTx9k9/rTxo3G+cjlWTC3m82eYO33y1y9OAJNHDyfd9as812g4+OkTz7Va1Kf/YeFca9PrwMAXDOvbi5NqNWfCwrgUFsaWT6F0a9eC8HDrxNOsVhXqm0y8Av4BQgB3rZa61W1VdRKPMYP6ke/ubZ5p1BwKC+O5VsvLY4dZuXyx3TFH9PekyP27PFGrORz2iWc6LU+897NujbW366FDB/DeuY2bwCXgPrDJbGZE/94J0mp/Ffj4XGb58qncuqXl+PEwbt8OZ9SoV3Tp0jheud7EYOjQAbx9+wQ/Pzh/3pKYZs2C8eP/xGiM/8xDUsFsNtOrVwu6dHnB/fvhHD36iYcPdXh5LeHEiaN2x+zRozm9ewdy9244R4584v59Hdu2zeP06eRljwGMHTuCt29v8eIFnD5tOW3bv7+ejh2tBRFNJhNdujRm9Oh/uX07nOPHw7h1S8vy5VOiqw7JiUQlepVKdUGlUtlHbUkieHvvpw4QW6exHNDQZObAAWvT6x3btzBbpyXqnlgKzNZp2bHdvtOMZrOZnTu3M1uni/aClQEztVp2bNloV0yj0ciefbutYiqAGVoNOzeus+rrtXsHbYxGYm+R1gQqAEeOHIpue//+Pe81auZioZ8CKLGoLYbYyRLRaDQcPn6UGZGRRN1nOwDTNBp2rYt7SjExCA8P4+TpU0zT66NjOgFTNBp2rFlh1Xfp3Fn0x0ITjUJTwN1sZunSBXaN/1/Bzp1rGDRIQyxpF7p2NSOTfeLqVfuEsI4d285ff0GWz+q9AoFFqtfFBebMmZEEs04Yfn4q3rwJZPBgE1GVjQwZYNQoNTt3rrIr5r17d4mIeEX//ubomJkywfDhanbs+H5D+K/h2LEtTJ8OUS6QAgGMGQMREWp8fX2i+/n6XkEu/0TXrjHXZssGAwdq2LVrbbLPMykKnVvd3d0FWPYIR6tUqm9y7fj8CPJVmEw60tsqAgHp9ZEYjTrSpo3xgw2LxXuOQjrgk1Zr1S+xMJlMROgj440ZptXYFTMyMhKtXo9tRT4d8EmjtoppNGhJp49b605vNGI2R0b3DQ9/ixmLebZtTAPYNc+PH40IAOd4YoapI74aM76/m80aJEIBtu98OiDCZu16dXg01z42MgAaTbhda/oepOR4kZFhpEtnXdYUCCB9eiEikcHOuRhIl866RSCwJKrQ0HfJur4nT4ykSSPG1gMnXTrQaD7ZNbZYbCRtWhG2JfF06UCnC0v298tg0JPWJjFIJJazBFptzJpEIgPp0wvjzDN9ejM63dfX/r3r+N5EX0mlUgW6u7vLgAVY9Kzaf0uAxNboS5WqSEeJhIkGQ3TSCQd2yOSsKl3RqoZVrUw51l++yJ+x9h/WCQRUL1ve7hpr5RKlWH/1CrG9hdYJBFQrX9HumGU8irDl9k1i/cizTiikWuWqVjFLl6vCaOVyhqsjiNJxfA8cALqWjFmTo2ManAQCdpjNVm/CGsBRJrdzniLyZs+Bl78fLWPPUySiSrWaX4yZcE1bTuaMmTjw/BlNYscUi6lSo7bVNRXqNGDl8kX0JubD+i8WEbTDzdumaM08pWv05crVY/36c7Rtq45Ojk+ewI0bkeTN62HXXDJkKMCKFXdo0YLopKNSwaNHsHJlv2RdX+bMuQgMNHHrFhQtamkzm2HdOjkVKtS3a+xs2dzx9zdw7x4UKhQTc/16BRUqNEj29ytXrmKsWnWFKrFIehcvQmiogNKlK0eP7+5emOvXI3nyBHJ9Fkg1mWD9eiUNGnx57V+o0Sca30WvjCrnqFQqHbAM4sirJxk8PApTp2kLyiqVLANWAOWUDlSp3zAO93rszHlMdXSkr1TKDqCvVMpUR0fGTLf/NOO4WfMZ7+hI/88xPWUy5jg5MWqq/Sc5J8xdyAgHBwZKpWwHespkLHFyZsTEaVb9KlasjEe16pRXOrAKC2WyrFJJq45dyZXLWpOny6Ch9AQGANuBDsAcYMqiZXbPc9L8pfRVOjBMImE70FkmZ5OLC0PGTrArnkAgYOrCZfRQKhkhtsTsIJezyyU1A0dab8qNGjWW0FSpKI3FaHweUBSoUKkKhWzOBvxqaNGiNTqdOzVrKtm4EWbOFFC5spLRo6fYrSGzefMurl8XUKUKbNoE06ZB2bJQtGjJZJfmlsvlTJmygN9/VzB5spDNm6F+fSVPn2alU6fudsVUKpVMnDiXmjUVTJ1qiVm3rpKXL3PQtm2nrwf4TixYsJojR8TUqwdbt8Jff1lOMnfrNhi5PEZe29k5FaNHT6FSJQUzZwrYuBFq1lSi0+WjefNWyT7PbxI1+8yyqa9Sqe65u7s7AGKVShX6uXQzBSigUqmafClGLGTnG1k3ZrOZkyeP471jG1KpiJqNWvD773XjpTK9eBHAhL9G4f/wAXnyF2DC5OlkzZotnqiJx6NHjxjUryevA56TMVduFi9fk+CXw89PxdGjhxAIhDRo0CjBfsHBQWxau4rgx4/IWagoHbr0IK3tsyCW8tGUKRPYtX0LIpGInr370adPXKEysGxgjh8xmMiPH3HKmIn5S1dTpkzZ71k6AQHP2bx2FS/8VBQqU5b2nbp+lV76tTvgBw/uM3n8aIICAihaugzjJ0whTZq4a9dqtfTr15NrJ/9GKJXQtmcfhtpITqcEkvKO/tOnUA4c2Me//76kVKkyVKlSLV5fX51Ox969uzlz5gCpUrnRunX3BAXNHjy4y8CB/Xn79g1VqlRjzpyF8dImQ0Je0bVrB168uIFIJKddu34MGzbqu9ZjNpvx8bnEhQvncHV1pXHj5vFSmsFiQjNv3iRCQ99QrlwdRo7867splnfu3GLbttVERLyjVKnatGjR5rv1ZB49esisWdP48OE9DRs2pVOnLvG+R69e/cvo0UN5+PASTk5pGDRoAnXr1os35vXrV9m5cy2hoe+oWrURTZu2+CprL8WsBN3d3Rdh2QPLALzFIgveAIvznejzvwfAnyqVytanISFk5xsTfWx86UsXEhJC87rVyfThAxUjwrng4Eiwiwtex06RPn3RtyfhAAAgAElEQVR8Fd+v49atmzSvUw0Pk4nqWHTJ/cRiDp66GEfLetGcmaxaNJfWBgMGgYBdIhHDxk2iS3dPu9YD0KJBbW5d8aETFi7rVqB+0xYsXpH8Gzn24ktrevzYn5b1a1NMp6VIRATHHBwQZMvBzkPHcXSMqUeazWbGjxiM985ttI6MJFQsZq9AwMyFy2nUpFlKLQVIukR/9+5t2ratR6VKBvLmVePt7UCaNB5s3HjQ6i7wWzBz5lSWLZtJrVqWEsb27fDxo4gLF/zivXGIQlKsyWg00rt3W+7fP0ezZmoCA+UcPSpgzZpdVKxY2aqvj89lunRpyu+/m8iWTYuXlxJ39/KsXLkzSc5GJNV7tGjRPObOnUCjRhYHqi1bwMEhE2fP3kuxMxxR+H/lGWuLL72h/bp2INuxw8yKpcI4XCwmoE49lqzbbNdkyxbIRcu3b5geq20AcOK3rJy7EWM/9+jRQ5rXrsJtrTZ6E/EZUFIm5x+fG2TObO0elZj17N27m1Ge3XiEZcMSLEYYhYHDpy78tCWML62paa0qtLhzK3ofxQy0k8lI392T0bFO8V64cI5h7VtyXa0mqlhxB6iqUHD1rl+SyOAmFkmRRMxmMzVqFGXkyGe0s1joYjRCw4YKSpUaSd++g745ptFoJG/e1KxZAy0/b6QYDFCjBrx7V4DTp30SvDYp1rRz5za2bh3MmTNqom5OT5yAbt1c8fV9HJ0YjUYjZcvmZdmyN/zx2XMy8v/YO++oprIuiv9CEkICNiwICnZQsTP2sTs2BMTee++9995Fxd4VHUfHAoq9txl7RcUOKqiAFCEkgSTfH1EgkCBG1PGb2WvNmsX1vvPufQmH+87dd28V1Ksno0WL+XTo0PmrxgGZM5/Y2FhKlbLD3x9q19a1xcdDxYpQrVpfZn1FudYU/Ktkir8E/scOMyqV1O7oxEQOHj1k5Ir0odFoCAoPY0yq9jHA81f6tEX/g750TEjQY4oUAjwEcOSIP6Zg8/o19CY5yQMURfdK5e2d1uDgn47IyPfcvn+PvikWGQJglFKJ/55den0P7tlF7xRJHnR/4KoKRZw5c+q7jDczERT0gvDwN7Rrl9wmFMKIEfEcOrTDpJg+PluRSHRuSZ8gEsG4cfDmzYOvHPHn4e+/ncGDk5M8QIMGYG2dwM2b15Pa7t27g1Qan5TkQSd+NnSoHH//72fk8jns2LGNggWTkzyAVAqjRsHJk7uMXfaPxv9lohcAqY+TqMFkR6JPSB3T0JEVgcAMjYH7aJJGZgIEAgwdZUkEgzXDfzo+fQ6p3+M0pP2MBAKBkeep/erP80dAIBAYlPzQnVUybT5mZoZjmnimyoT7mxm8l1qt/xkJBAI0Gm2asarVX/+7mZkwMxMYnI9G888a55fg58sSGYB7EzfmiERJiUQLzBGJcTfRB9bMzIxCNnmZlaJNi273uUgqG7Cmbh5sF4lIKfT7GPDTQpMmpt2/Z58BrANSCvg+BA4BAwd++av+j0b27DmoUKYcy1P8kdIAcyUSmqZSZXRv1ZY1Uikpfa+vA5fVaurUqfddxpuZcHAogK1tfjZvTm5LSIAFC6S4uZnmWtW+fSdUKl0d+RNUKpg1C+zsSn3dgDMAN7dOeHnJ9Ay6Dx6EmBgLPUZcqVJlSEiwwjeFt4xCAUuWyHBz+/qyTWahffvOvHwp4HgKb5m4OJg3Dxo0aGf8wn8w/i8T/cTZCziR34HqVlaMFgqpbmnF8fz5mTh7gckx1+3cywahEBdgJDqK3y6xmA079X1GHR2dGDR6AuUsLOhrbk4PiYTKEgumzJ6PrW1qBfaMwd29GZVq1sYJ6IWOMlkBaN2hMyVL/pwysPNXrmN5zpw0sLRilFBIOUtLQks4M2iYvpl1lSrVcO/cjRIWFgwSm9NJKuU3CwuWrFyvt2n7s0AgEODltZWJE7Pj5mbJyJEiSpe2RCCoSPfupplEC4VCBg2aTJ8+upLJsGFQuDA8fChi717TypVfAk/PlhQp0pASJWQMGyaiZUsZXbtasnLlDj3Wj5mZGStWbKd37yy0aGHJ8OEiSpSQkS9f3X+U7LJMJmPs2Nm4u4OHBwweDIUKgZmZA5MmTf/RwzMJP9VmrEajwd/fjwO/+2AuFlLfszXu7p4GyxeJiYkcP36UR48eUqyYEw0aNDK4W56YmMjUqRM5uecPQECD1m2ZNGm6wb5yuZyFC+dy794dypWrwPDho42yJF68eM7hw/4IhWa4urob3YR98eI5m9es5NVHemW33v0M/kFQKBT07NmJ86dO6lQ3m7fCy2vFV5VuoqKiaNKkHu+fPAaBAMfKVdi719/g3J88eczmNSsIfvyIUpWq0LVnX/KkPmKZCp/bGFMoFBw6dICXL4MpW7Y8NWvWNjgflUrFsmVLOLp7J+YyKb2GjKRZs+ZfPmETcf9+AFu2rOD9+1c4O1enc+eeBqmlCoWCP/7YzpkzvlhZZaNNm15pWCefcPXqFRYsmMT796GUL1+L8eOnGVQu/RI8f/6MoUMH8PbtGxo0aMSUKTM/q0qZmZTRGzeuceHCeaytrXF3b2Z0o/zDhxgOHPAlPDycatWq4+JS8atLIlevXmb79jXI5e+pWLEB7dp1xsrqyw4VpYav714WLJiMUvmBSpUasGDB0u+uign/QtbNiP69uHPoIIPlcZgBy2WWFKv/G8vWbTHpi6LRaKhZwRlhyGtGoyvHzAWE9g6cvX7vM1d/PW7cuEbHFm70UKqonJjAaXNzdllYsPfwKT0H+8TERKqVdiRbRDjD0RmEzAayOxbn6AXTNE8UCgWlHPJQFBgGyD/GjJVIePAyTK/vX39dpEe7FvRVqaiQmMgxiQQ/qQzf42fTdbDPjCSSmJhIh2ZN0N67Qy+5nGhgkUxGiz4DGDlu0lfFzgiOHz/C0KFdGTRIibOzGj8/CWfOZOPgwYvY2Ngk9VMqlbRu/RtZsjyiWzc5YWGwaJGMrl1HMWDACL2Yhw4dYPTongwZosTJScP+/RZcupSDAwcupEuF/Bb4EYqcmQ0fn40sXDiO4cMVODho2b5dyosX+dm376zJyX7TpjUsWzaZESMU5MunZds2KSEhBdm79/R3T/b/qkR/69YNeng04UG8nE9HK+IBZ5kM792+VDTB8Wfr1k0sHDmEZ+gExQDi0D29Kd6rv7kMrkfd6vS7d5eUldkFAgEX6/7Gut//TGpbtmwx22dO5QE6MTXQyfUWAFb47KJBA32Z5oygWbMmhF26QAAkiapFfow5dNJ0Bg0aCujogL9VqcDU50/15AqmmpnxyK0ZS9dtNnqPzEgivr572TB0ABfi4pIE0N4BxSUSzly5bXI5LCPQaDRUqeLIpk3v9BgYQ4aIUam6Mn36oqS233/3wdd3JMePJ8sVvHoFpUtb8PffD5LeANRqNRUrFuGPP95TrVpyzH79zBGLezFpUkoC77fHz57o4+LiqFChCH/9Jcfpo0WAVgstWkgpX34iffsO+uKYHz7E4OJSlGvXFElyBVqtjgJbvfp0evQwrcRmKv5V9Mpz587SIkFFyvNzUqCVQsHZs6dNirl//x46kZzkQafM2A7Ys2e3yWPNCJRKJdfuB5C6MtlFq+XsxfN6bYcP+NKN5CQPkA3wBHbu3G7S/e9fu0oPkpM86MTQmgKbNiWrUkZHR/HsVTAeqa7vrNFw9jvQG88dPUzHFEkedDTTuiIRF1M9p8xGcHAQCQmxejomAJ07J3Du3BH9cZ47QKdOcj3Brvz5oUoVsZ4M7bNnTxGLlXpJXhdTxblz376e/v+G27dv4uQkSkryoNPw6dw5nnPn/EyKef36NcqUMU9K8pkR80fjp0n02bNnJ9SA61OouYQcOVLrNWYMOXJYY0i89+XHf/uWEIlESMViwlK1hwDZUr1uZsuRw+A4XwG5PumjfiHMJBIM6U6/BKytk+duYSEFgYDIVP1CgexZU2taZj6y5czFawN7BqECswy5XH0NsmTJSmxsIql92UNDSXPvbNlyERKi/+uk1UJoqFbv+5k1a1aiohJJ7cuui2na9/jfjGzZshMampiGDhkSovtMTEH27NkJDVWnoYG+fi0ge3bTYv5o/DSJ3s3NgxMCASdStJ0FDgrAw8O0o/BjxkzA72OcTzgJHPn4b98SQqGQ1q3bMUwiQfmxLRYYZSGlfTd9gadRYyawA51P7iccBC4CI0aYpvkyYuwENgHXUrTtR0ddXLcuWWPfwsICD1d3Rpqb80kkOhoYK5XSrue3f4Vt07EL60Ri7n78WYtO/iHY3Jxatep+03vnzJmTX3/9lfHjxXw6fxceDpMmyWjbtr9e37Zte7B8uYTAwI/j1MK6dQJiY7NQuXLVpH42NnmpWLEikyeL+eTz8e4dTJkio127Ad90Pv+PKFnSGWvr/CxcaJaUmIOCYP58Ge3bG5ccSQ9ly5bH3DwPS5cmx3z2DBYtktKu3fct22QWfppEnyOHNeu2/UHnbNn5JUsWKmXNSpusWVmz5XeTV7WOjk6MmDiVJoAzUBLwQMDYabO/uZIfwMQZc4mvXgMHCwsaZMtGAYmEPI1dGTh0pF4/F5eK9Bk2mrpAacAJaCsQMGPhUr0NwS9Br179cCpXgRroTpoWBToAzdp1TDP36Qu9CK1YGQeplN+yZqWQRIKjZ0t69u5vIHLmwsmpODMWL6OWVMavWbJS2sqKyXnzsvVPP8Ri8ecDfCUWLtzA9etlKFRIRv362ShaVELNmt1o00Z//6ZcuQqMGTOXqlWl1KyZlZIlrVi8OD9btvilYRItWbKFixedKVRIRr16WXFyktCgQS88PVt+8/n8v0EgELBu3R62by+Mo6MldepkpVw5C3r2HEeNGrU+H8BIzI0b97F+fQGcnCypVSsrLi4WDBgwhSpVqn0+wD8QP81m7CckJCRw9eplsmWTUqxYaaMm3l+C2NhYtm/fgkBgRseOXb56Vz0mJhqv+XM4tH8PQjMz3Fq3Y/Dw0UbjPnv2lMjIN9jYOJA/v306cWPw8dmMRGJBhw6dTRbASong4CDGjBmBlZUlS5asMMpSOHLkEEtmTubl6xBKlSzF2BmzjaoofsLnNvq6dGnP5SP+qLRaLEQiug0bZVRFUS6Xc/XqZWQyGS4uFb/6RPDZs6cYNKgTcXEfMDODvHmLcPjweYPz37fvT1aunMm7d2GULVuBMWPm4Oxs+CBSbGws165dIUuWLJQv72JwnFqtlj17drFixSzevg2nYsVKjB07mxIlvs+ZiODgIBYvnsaFC6fJmTM7rVv3oVu33l/1TF+8eM7ixdO4dOkMOXJkp0OHgXTu3P27ndzWarXcuXMLjUZB4cIlyJbt68t6Wq2WW7duEBUVhYvLL1+tq6RWq9mwYTU7d64mJiaWWrV+Y/jwyUap15/wr2LdpMY/lS2QkJCAW93qlHr+jBEqFYnAbImEd86l+fPwSaM00H/qfAB27dzO/DHDWRIfTwXgGDBBKmPb3gO4uFQ0el16c3JrWIegm9dZje6N4jAwChg0ZgIjRqRWFcpcPHx4n8aNq9CzJ/TtC+/fw+jR8PixBffu6fsPr1u3nC1bZrF4sZySJcHXV8CsWTL27j2VRrU0o1ixYhG7di1g0SIdU2TfPgFz58rw8ztH0aLFPh/gK/Du3TsaNKhI9+7RdO6sISQExo6VUbZsW2bMMM2a8c2bUBo2rEzv3jF07Kjh5UsYM0ZG5cqdmDzZ9EOKpuCf/Hs0Zkw/Hj3ay5w58djYwKZNQnx8snP8+DWjks7wX6L/R36gBw/6sW5QHy7GxSUpl6iBMpaWzNz2h9EDNP/U+Wg0Gio5F2VXRDgpCazrgD3VfmXbfuNMEWNzio+Pp0QBG86Dng/uCmCqSMT9kPdprslM1KlTmaJFH7B/f8ox6fxLp03zTlJRVCqVlC9fiAsXYvVYHfPmmXHjhhve3l+uhBofH0/58oW4ckWux+qYMcOMwMAWLFnybWWn58+fzocPy1i9OtmaMjISCheWcP78XZNkvGfNmoBavYbly5NjhodDsWISLl16YHJp1RT8U3+PXr9+Rb165Xn+XElKDkO3bhbY2Y1g6FDji5t/Fb3yZ8GdWzdonCLJg06s/zelijt3bv+oYZmMqKhIYj7EkPqUQiPgzr27hi75LK5c+Qsh+kn+U0xlKtXRb4GIiGc0a6bfJpXCr7/qTOg/4fXrl1hZafWSPEDjxhru3r2GKQgODiJXLjO9JJ8c86pJMb8Ed+9eonFjff/hHDmgQgUJ9+8HmBTzzp2/0sTMlQtKlZIQGPjt1TN/Bty7d5dKlSSkJqo1bqzg7t1L3/z+/yX6TIZDwULcNFCLv20hwd7e4QeM6OuQJUtWzEQinqdqvw3Y2+UzKWbx4iVJgDT0ztuA+DuoA0ok1lxNlVM1Grh9G8qUKZfUliePDZGRiYSl4sDevg329gVNureNjQ1v3yYQmYqvqotZyKSYXwJ7e0du3tSnqyYkQEBAAg4Opn0/HRyKceuWfipRKuHRI1W6e07/Jjg4FODevURSr2Nu3RJhb+9o+KJMxH+JPpPRrFlzLkskeAsEqNC5Qc0xMyPI0oqGDRv/6OF9McRiMd169KGLVEbQx7abwBCpjD6jTKN22tjkxSZnLtqSnOyvAf2B6k1TH83KfMyYsZhNm2DnTp1EbkwMDBkCHz4I9Gi1VlZZaNWqLV26SAkJ0VEmL12CceOk9Oo12qR7Z8+eg2bNmtOliwVv3uhinj8PkyZJ6dVr1OcDfCW6dh2At7eYAwd0f9wiI6FvX3PKlnVJ4z+cUXTrNpglSyQcOqSbT0QE9O4t4ZdfqqYrkfFvQokSJSlatDQDB5oTFaV79vv2wdq15nTu/O0pm//XiV6tVvP+fQRqtSE1928DK6ss7D54nF1ly5NTLCaXSMTJipXZ7X/cKEMoMTGRiIjMHadCoeDp0yckZqAUEhLymrDUy9YUGDl+MpV69qa8VEZuiQVu2XMwcPps3NyaGb3mczh99Q6vc+XGEbACagNFKlZiw4atRq+JiYlGLpebfM9PaNTIlV69htGvH1hZQe7csHOnOTt3Hk8jAjZ16iIKFuyAs7MFuXJJaN8+N5Mne1OrVh2j8e/fv0toaIjRf585cxm2tu1wcpJgbW1Oly55mDlzNdWq/frVc1MqlURGvsfY3pujoxNr1+5i7NgC5MxpjoODOXJ5E1at2mlyzJIlnVmxYgcjRthjbW1OoULmqNVueHsbNxP5XMxP0Gq1REa+R6lUptvvZ8C6dbuJiGiAvb051tbmTJlSiA0b/qRw4aLf/N7/l5uxWq2W9au9Wb5oPgpFPBYWUgYOH0WvfoO+i3HA06ePGTeoH5c/uuvUqlqd2ctWpXmN1Wq1eHstZPVyLxISVFhKZQwdO5Eu3XuZfO/ExEQ6tnTnr0sXABAKBDRt3oplq9an6Xv4sD9j+nbnfXw8WiBfDms27vEzak2oUqmIjo7G2tr6s6qIkLGNsbCwMAIDH+Li8otRM+c7d24xYUJ/7t59iEAgoEGD+syevTJdpkJGce3aFXLmzPXZcxMKhQKJRItWKzFKGZw5cwpbty5BLtetbLNnt2D37pOULFlar59KpWLOnAn4+GxBo1GTN29uJkxYQJMmbibPQ6lUMnPmGHbu3IFGo8bOLg+TJi2mQQPDb5FarZaIiAgKFLAhLs7wAkOhUDB9+ih27dqJVqshf/68TJ68hHr1GqQbUyaTGaUSy+Vypk0byZ49u9FqNTg45GPKFC9q1057+O306ZNMnz6M4ODXCARmtGjRiilTFqZLf/6nbsamRFxcHAqFAmtr6wzlo/82Y41g6+YN7Jg7i+Mx0USqVByPiWbn/Nls2ZQ22WU2YmNjaeXaAPfrV4lUqwlXq6n210Vau/6GSqW/YbVquRf+SxdxPvYDkUolB6IiWTttErt3G19dfQ6dWjcj/NIFbqNTpDyj1XJuzy4mTNAvNTx//ox+XdoxKT6eD+jc3ltHvqd5g9ppxvkJ5ubm5M6dO0NJPqPInTs3v/5aw2iSf/v2De3aNaFv33tERSXy+nUCDg4n6NTJ9bOrwYzgl18qZehwnIWFBXny5DGa5H1997J+/RJWrgS5HN6+BTc3Bc2a1UjTd9KkoTx/vpWAAAXR0QmsWRPC2LE9+ftv0zflxo0bQEjIDu7f18Vcvvw1w4d35do1w+qmAoGAXLlypZs0R4/uTUTEHwQGKomOTmDx4pcMHtyJW7dumBxz+PDuxMX9yaNHSqKiEpg37wX9+7fjXqqN/bt37zBgQHvmzXtBdHQCjx4piY39k+HDu2fgafyzYWlpSc6cOb+rW9X/5Yq+SqlibH/3liop2i4D7fPY8Pe9x6aMNcPYvn0rZyaMYb88Tq+9lpUVXZetpmlTd0C3+innWIBj0VGkPHpzEhhWsBAnr3w5Q0elUlEsfy5uAMVTtB8BOovFBLyOSGrr2rkdsiP+pJRu06I7Idxs3CSGDfv6enFmrK68vOYRHr6QNWuSX921WihZ0pIFC/ZRuXKVdK7OXKQ3HxeXInh6hrE0hYWvWg22ttCnzzQGD9Y5gUVHR/HLL8V4+lRJyheStWvhwIG6bNy4ny9FREQEVasW5/lzJSkleLy9BZw61ZA1a4z7nBqb09u3b6lZsxRBQUqypPB3WbJEwN9/p1+WMYaQkNfUrVuOly+VpPxbMG+eGXfv6lNLhw7tTpkyexkzJlnERi4He3sJp07dws4IEeBnWNF/Kf5b0RtBUNi7NNS9Ch/bv/m9nz/DJVWSB6igUBIcHJT0s1KpJPxDDM6p+wFB6dR300N4eBhqdBIJqWPGJiTotYU8e0LVVP0EQGXg4cN/DiUuKCgQFxf9+qxAAOXLQ3Dwix8zKANISIiiUiX9NqEQypaF69eTKT5v3rzBxkZM6qqTiwsEBz8z6d4hIa/Jn9+c1BpvLi5aXr58alLM169fUqiQRC/Jf4oZHPzEpJgvX76kWDEJqRf8Li4aXr58nKrvY1xc9JXKZDIdN//Vq1f8hy/D/2WiL124CMdTtR372P7N7122HEctLfWMrzXACYk5pUol12olEgmF8ubVE1RLGqdjcUxB3ry2WAgE/JWq/RiQQ2ap1+ZcsTKpBVcTgaNAjRqGD3X9CJQqVZmjR/Uzg0oFZ85ojO4l/AhkyZIff3/9Nrkc/v5bZwX5Cfb2DoSFaXieiq969KgZpUq5YAoKFSrM69eJvEzFVz16VIizc/oyFcZQpEhRnj1TERqaOqaIUqUqGb7oMyhWrBiBgco0dNWjR8U4O+vHdHauxLFj+lpG797Bw4dKihb99puX/28QTp069UfdOzswND5eZdDB/nOwtJQglxuuJectUICBhw9in5hIDnTH6/tLpUxdvNxkCllGUbhwETb9+QfXP8RSTKMmFBhqLkFVvDijJ0xNqssJBAJy2tgy5OQxCiYmkhXwA4ZZSJm1fJVJtDSBQMDrN2+YcecWjkAWYB8wABgxbZaeNk3lylWYsWo5L7RaiqGjOfYB3mbJive6zZlSP0zvM8ooihVzZMGCrQQHKyhSRMOzZ9CnjxQHh5p07/591R7Tm0/58hWZPn0LHz7o/FofPYJOneDDByne3puT+onFYrRaNVOmXMfRMQGxGDZv1qktLlq0npw5v/wUqbm5OSqVgunTb+HomIBIBBs2CFi8WMqSJevTldw2NicLCwvk8g/MmnUHR8cEhEJYu1bA8uUyFi9eZ5KkslQqIyoqgvnz7+PkpHvDXL1awJo1upjZsiVryRQtWoJRo7agVidgbw937kDXrjJcXbvTqJH7F8/nZ0bqOQkEAmQyc4ClQFRGYvxfJvoiRYpSvFwFvO4HMCsulkfFHJnutYL69Q2zBb4EL18Gc/v2TcRiMVmypNVjFwqFuLdozaX37xn95BHbJBIqde7GguVrkEgken2LlyhJIedSLHv0iHmKeF6VLsvMpSuNyiRkBL81bEykSsXcWzdZrE7kL6ssjJg2m56pJIWlUimN3D1Zc+4M8yMj2WpmRhaXiuw7eirTrNIy45dOIpHg5taKkyfDmDz5FX5+OahTpw/Tpi3K1E3hz+HJk8c8fvwAkcjC4POxtbWjWLFSLFhwhGXLEti+HczMCnPmzO00tNpKlapjYWHL/PkPWbJEjVJZg8WLN36VqFnVqjURiXIzf34gXl5q1OpaeHltxtExdSFPhzdvQtm2bQtv3oRQoEARg3/Yq1evjUZjzbRp91i8WI1aXRNv721G9XhCQl7j47OV0NAQihVzNBizRo16qFRZmTEjgGXLNEBtvL190myIZ8uWjQYN3Nm58zmTJ7/hxInctG8/jiFDxhiMq9Fo8PPbx5kzp7CxscsU43itVsu9e3d5/DgQa2trzM0lRvtGRUVy9eoVFApFpks+ZEai/+xmrJOT00KgBbrN09KBgYH3PrY7AluAnOhIG50DAwO/ZKezIN9oM1alUjFqQG+OHvGntETCXaWSBg2bsHDlOpPVLpVKJcP7dufUieOUkphzV6mksas785evTiOXu2zZYpbMnEZ+tCQCbwUCpsxfQpcuxhkD/4ZNpJ8RkZHv6du3DQ8f3sHR0ZxbtxT06NGXMWOmf1fWRGaif/+uHDy4FycnHTsoIUHMpk0H0kjwvn37hl69WvL69RMKFBBx756KIUPGpPHABejVqwNHjx6geHGd6YdGY86WLf5pLD5DQ0Po2bMFb98+x95eRECAihEjJtCnz5A0MdesWcqiRbNwdjbn5ctEbGwKsX79njT2kVevXqZLF1eEQhW2tvDwITRs6Ma6daa5rwEEBb2gV68WREeHYGsr5MGDBMaPn0mXLmmpz0uXzmXFisWULm3Os2cJFCxYnLVr/8w0/9/vtRm7H6gJBKVqXw2sCAwMdESnR7UmIzf8Hlg0ZwbRRw8TrFRyJiaGYKWSmGNHWDRnhskx502fhPLkcYKVCs7ExJqDObUAACAASURBVBCkVBJ26CBLF83T6xcQcI9FM6fij5aHwBNgp1bL5FFDCQl5/VXz+g/fHyNH9qRkyRsEBcVz5kw0gYFKTpxYx59//vGjh2YSNm5cx+nTe7lzB27dgtevYfToBLp2dUOTyqapf/921Kt3n+fP5Zw/H8OdOwp8fBZw/Li+jeKqVd5cunSAgAC4eVPnljV0qIouXVzTxOzTpzWurg95/lzOhQsx3LypYMOGOZw+fVKv3+nTJ9mwYQ63bim4cCGG58/luLo+pG/f1nr9NBoNnTu7MnSoipAQ3f0DAuDSpQOsWbPCpGek1Wrp0aM5HTo85cmTOC5ejOHy5Xi8vCZx+fLfen39/Q+we7cX9+4pOHcuhqCgeKpVu8uQIZ2MRP8x+GyiDwwMvBAYGKi3zePk5JQHHZnj949NvwMVnJycvq+FvRFs37qRJYr4JH9ZS2CJIp7tWzeZFE+r1eLjs5WlCgWfXtqtgMWKeLZvXKfXd968mbQAUloeuH78ecGC72v8/B++DhEREZw7d45583S1dAAbG5gxQ86OHd4/dnAmYtOmxUycCMU+Vl/MzGDUKLCwSGDfvmSy7YsXzwkMDGDKlEQ+Vcjs7WHSJDnbt6/Ui+njs5QpU3R7E59ijhsHAoGKw4eTd6ifPHlMcPBjJk5UJ3nrFiwI48fL2bFjlV7MHTtWMmGCnAIFkmNOmKDmxYvHPHmSXDg4dOgAZmYqxo0jKWbhwjBlCmzduhRTcOfOLeTyUEaM0CTFLFYMRoyI5/ffV+v13b59OdOmycn3ke0pFMKMGYncunXjH7WwS2vGmTHYA68DAwPVAIGBgWonJ6eQj+3Gz9IbwMdXEJOQO7fhOlykXE5qlm1+4H1crNFr0oNGoyFGEY9dqvZ8QGSqmLExURjiORQEQj5EpXt/U8b2T8fPPKeYmHdkzSrC0lJ/nyF/fh0f/mecm0oVR/5UPhcCgY7vHxUVnjSnoCAVefOaIxbrm9vmzw8fPrzXm7tKFYd9Ku0ynZkLREeHJfV98kSJra0YoVDfhDd/foiN1Y8ZGxuZlDw/QSQCOzsxoEzqGxMTQd68kPocW/78unGZ8hlptUry5ROSujKXP7+WM2f0x/nhQ1SacUokkCuXGIFAlWnfka+NY2qizzR8ixp9jQq/sOPqZVJW07YDNV0qmlwzrl62HDtu3aRrijYfoEalKnoxf2vYlA2XLjER+LR1EwfsBma7ehq9//9DPTs1fvY5WVrmxMzMkvPn5dRIccDVx0dEtWq//ZRzK1GiOhs3+tOiBUmJ7PlzXblj/Xr3pDnlzVuQ0FANd+5AmRQsVh8fCVWqNNKbu5NTFTZuPIGbW3LMx491/9Wr55rU186uMM+fJ/LgAZQokTKmRZqYlSs3ZPv2uzRunHyG4v59eP48kXz5iiT1rVfPlbFjh/L4cfJbilYLGzeCo2MVkz6jwoVLcPu2iufPoVCh5Jg+PlIqV9YfZ9WqDfHxeUaNGsmLgevXISrKjFy58mfKdySdGn2GYSqP/iWQz8nJSQjw8f92pFWe/SGYMHch4y0tGSMScxAYIxIz3tKS8XNMd7uZOG8xI2UyxotEHARGicVMt7Ji7Ez9Gn2fPv0R5spFFWAHut3qXwAbhwJ4eppmYv4ffgyEQiHTpy+jRQspc+aY4ecHPXpI+PPP7AwcaJp65Y/GvHleXL4sxtUV9u6F5cuhcmVo2LAp+fIlL8slEglTpiygYUMpixYJ8PWF9u0tuHQpNz176nsFz5+/nLNnRXh46BQZly6FqlWhadPmep7GMpmMyZPnUr++FC8vAfv3Q5s2Fty4YUPXrr31Ynbr1ofr1/PSpo0F+/eDl5eA336TMXnyPD25DBsbG5o08aRaNd199+0DDw84f17EggXLTHpGWbNmY/ToKdSqJWP5ct1z8vSUEhSUn/btu+j17d17CCdP5qRTJwm+vrBggQBXVynTpi3+Lp7GGUWGJRCcnJxeAE1TsG7OAOsDAwN9nJycOgI9AgMDjUv6pUVBvqHDVFDQCzau9ubxndsUK1OWbn0GULCgYb3v69evsnLlcgQCAQMGDKZ8ecMHV549e8qm1d48DbiHU7nydOszAAeHAmn6KRQKhg4dwKVDBxGYCajXvCXz53shEhl/gcrI6jco6AXHjh1GJBLTpImbUWPwhIQEjh07wosXz3F2LkXNmrWNarQ8f/6MEyeOIhab06SJG3ny5El3DF+CH7GiVyqVHDniz6tXryhfvgJVq1Y3yo559CiQ06dPIJNZ0rSpu1G++fXrV5g1axLh4aFUqFCdCROmZxqj4kfg7du39OvXjcePr2JuLqVz56EMGTLcYN9r167g47OSsLAQKlf+jS5dehn0Yw0JeU2nTq158uQ+YrE5AwaMNCqjcebMaRYtmkZkZAQ1azZk7NhJBv1YY2KimTt3BmfPHsHaOhcjR06lVq3aBmOuXLmMVasWoFIpcHIqz9q1W8ib1zbjD8UALl48z44dq4mODqdGDTc6dOhq0FM4MvI9W7as4+rVk+TOnZ/Onfsb9VOOjo7C3/8AMTHR1KxZh5IlU5+NT4vvYiXo5OS0DGgO5AXCgYjAwEBnJyen4ugWrDmASHT0ysCM3PQjCvIPsBLs27MLh/320QLdCdZ9gHvzVixfbbql2/rVK1g4axrNBJCIgANomTZvMW3bdTR6zefms2r5EpYtmEMzrRaVmRkHtVpmL15Oi5Zt9PqFhobQqkl98kRH8YtSySlzCbLCRdjheygNt3jZ4vmsWrKQ5gKQC8zw12pYsGwVHs0y583jeyf6Z8+e0rp1A4oVi6dUKSVHjpiTP39ZNm3ar2ekrtVqmTVrPDt3bqB5cy3v3ws5flzLqlU+1K37m17MJ08e07p1Q0qUUODsrMTf35xChVzYsGFPmnMRPwO0Wi1jxvTn2LG9eHqqCQ0Vc/asgE2b9lClSmpRjIwhMTGRunXLER4eTKtW8OwZnDsHM2YspXPnbnp9b968TqdOHlSurKFAARW+viKqVWuAl9cmvXMRarWaoUO7cenSMTw8EgkKMufyZTO2bfNNsxC7evUyXbs2p1o1DQ4OCezbJ6R2bTcWLlz73czJM4KLF8/To0dr6tQBG5sE9u4V4ubWhpkzl6ZL1f3PM/Yrk8jFi+fp6OnKXXRPDHR0yDLAn/7H03CAM4InTx7jUbc61xQKPvn1BAJVJRacvXrb6Cojvfk8eHCfVg3rcF0Rz6d9tADgV4kFF28E6K0uu7f2pNz5M8z4qG2vATqbm5O9S3emzJqf1O/u3dt0bNqA6/HxfBrRHaCWhQWXbz9M9zRlRvG9E33z5rXw9LzN8OE6Sl9iInh6WlCmjL4n54UL5xg5sjVXr8qx/jjNS5fA3d2S69ef6h2IcnOrRseOAQwcqPuOJiSAu7uUypXHMmDAsO82t8zCgQO+eHn14eJFeZKOzeHD0Lu3NVeuPEn3rdMYxo8fxaVLa7hyBSw/Ut18faFTJwEPHoQlnV3RaDRUr16CuXNDadlS1y8+HmrVktGlyxJat26XFPOPP3awbdtwzpyR86lSs3s3jBtny8WLD5ISuFqtpnJlR5YvD8P944HZuDioWdOSPn2W4+nZ8ssf0jeASqXil1+K4OMTTb16uraYGKhSRcaECZuMyknDf6JmX40VK5bRgeQkD1AUaA14e3uZFPOA3z7aJyaS0pTNCXATwKFDB02K6bd/D10SVKQkSzgDjYRmHDmSTF9TKBScuHCW0SkMTMyACSoVvql43757dtNNqSTln50yQF2hkKNHD5s0zh+JsLAw7t0LYODAZN62SATjxinw89NXWvT13U7//slJHqBaNShTxoxz584ktYWGhvDkyWP69k1eiIjFMGZMPH5+X24M/k+An982hg2T64mVNW4MuXIlcO2aaZ61J078wbhxyUkedHXynDm1epLbAQF3MTOLoUWKF0apFEaMkOPnt0Uv5oEDWxkxIjnJA7RsCQJBDAEByZLGt27dIEuW+KQkD7pxDBsWh5+fcROb742//76Eg4MmKckDZM0KgwfL8fX9ciXQL8W/OtGr1YkGaUdiQK3WGPiXjMRUIzLwliTSgkZjmoOUxkhMsVar50ql1WpBm5ZKJQbUqQ6uaNRqDG0VibWkOeTyM0Cj0SAQCNLQ7ESitPNRq9UY2icTi9F7nhqNBjOztDFT9/uZoNGoMbRo1z0nU+ekNfo8UzqcqdVqRCJBGtqioeepViemGadAAGKxQK+vWq1BJEpb9hCJ/lmfke47l3acurl/3gXua/GvTvS9evXDB0gp0Pca2ImOPWMKmri64yM2J6UgcjC648UNGzYxLaabB1skEsJTtD0HDmq0NGqUHFMqlVKjUmW8U2QmLbBQLKZJKtu/Jh6ebJJY8D5F2xPgqFpN/foNTRrnj4SNjQ1FixZh8+bkNo0GFi+W0Lix/mlKV9fWrF5tSWxsctvt23D1aqLeZp+dXT7y5bPHJ8WCSxfTgiZN2n6biXxjNGnSDm9vGSmd+c6fh5cvBfzyi2mqlNWrezB/vk5V9BNOnYKQEAFt2rRPaitduiyxsRYcPZrcLyEBli6V0aRJe1KiceP2LFsmI6W69tGjEBtrQenSZZPaypevQFiYiNOnk/upVLB8edqYPxJVq1YnMFDD5cvJbfHxsGKFJa6u7YxfmEn44Tz6H4n69RtQqUYtSpw/S3dADWwGfq1Tjxo1aqV/sRGULOlM134DKbvamw4qFYkCATtEIkaOn4y9vcPnAxhA2bLladezD2XWr6GDSoVSIOB3kYhxU2amqfnP9FpBiyb1OR8fT0V5HCcsrXifJw9/Tpyq1++XXyrh0akrZbZtor1SiVwoZKdQyJSZ8zKVefM9sWDBBtq2bcTRowmULi3H398SKMycOfqMkrp1f+PIEXdKl/alXbt43r8Xs3u3GQsWrNLbsBYIBCxatIn27Zvg759AyZLxHDxoiVhcjEWLBn/n2WUOmjVrwbFjeylT5iytW8cREmLB/v1mrFy5yWQdqDlzFlKjhj+OjhF06qTj5e/dCyNHTtXbBBcKhSxbtoUOHVrj6qqmQAEFe/ZY4uBQkbZt9YkK7dp14sSJ/ZQvf5XmzeMICrLA31/I+vVb9DZtxWIxS5dupmXLdri7a7C3V7B7tyXFilWjZSqiwo+EhYUFXl4baNKkK56eGmxslOzaJaNChfq4uhpX48ws/Ks3Y0FX7pg3bxY7t2wEgYCO3XsxYoRhhbwviblx41p+37IRgZmQbr370a5dxy/aWU8NjUbD+vVr+GPrJoRiET36DqJNG8MrgSdPHjN90jhCnj3FsVx5pkyfjY1NXoN97969zeFDB5FIJLh7NM+QrV5GkVmfUXx8PN27d+LevXOYmYlo3rwHU6YY1i26d+8Os2ZNJCzsJaVLV2Xy5JkGN5a1Wi3Xr1/l5MljWFpa4enZknz58huIqDsSP3v2JKKiQnB2rsbkyTMMUgy/BVQqFfv37+HcuYNkyWJNu3Y9KFOm3FfFVCgULFw4l7Nn/ciRIzcDBoxN1+w8I9BoNKxe7c2RIwfIkSMXY8ZMomRJw2qc4eHh7N27i/fvw6hatSY1a9Y2+Luh1Wo5d+4Mly6dI2fO3DRv3tqoMuS7d+/Yt283SuUHypevyq+/1vxuonNxcXHs3v07V66cIE8eezp06EWxYo4G+546dZylS+cgl0dTt64Hw4eP+Sx76z/WTSYkkTFD+vO37z76y+PQAitkltRu3ZYZ85eYFE+r1TKify9uHvannzyOBGCFTEajDp2ZnIL1khqfMzsf3Ksr908co688DiXgLZPh0a0n46bM1Ot7+/ZN2jVrQruEBCqqVBy3sOC0pSW+x86a/EZhKjLjM4qPj6d8+XzY2SUyaBBERcH8+WBjU4QzZ27q9b106QI9erSkW7cESpdO4OBBKTdu5MDP77zJvPdz587Qp09bevRIwNk5AV9fGffuWePrez5TzMnTg1KppH37xsADunSJ480bM7y9JYwePYeOHU3zTlUoFLRt2xBz80A6dZITEmLG8uUSJk5clGZV/TPiezO9oqOj8PSsTcGCb2jZUs6jRyLWrhXj5bU5DZPGx2cj8+ePY+BAJXnzatiyxRKBoCTbtx9KN9n/l+i/8gO9efM6PZu5EhAv59MLezRQQiplu/8JPUeojOLKlcsMau3BXbk8SVQtEihhIWXPiXNG9cHTm8/Fi+cZ3aE1t+RxSaJqEUAJCwsOnPmLwimcszzqVqfPvbukPL83ycyMJ009WLZen9nwrZEZn1GXLu158eIgt2+TtOH3+rXuuLuPz4GkEptWq6VWrVLMm/dSj4ExcKAYtboL06cv/uJ7a7VaqlcvzvLloTRqlNzep4855uY9mTx57tdM7bPYsWMb+/eP4uRJedKG8KNHUKWKlGvXHhv0Q/gcNm9ez/HjEzhyJD4p5v378OuvMm7ceIqlpWX6Af7h+N6JfuHCWbx+7cW2bcqkTeZz56Bjx5xcufIkqcz04UMMLi5FuXJFkSTVoFZD/foyPD0X0u4Lztj8R6/8Qpw+fZI2SgUpjxFlA1olJHLq1AmTYp46eYy28cnKmaA7UeapUXPqVGqDwwzGPH6E9vFyUtpd5ATcgdOnk8cZFxfHzQf3Sb0F1Uuj4aSJ8/nRuHnzNH37osfqyJcPatWCpUuTk3doaAgREWG4uelf36tXAqdOpfL4yyCCg4OIi4uiYaq96Z49VZw6ZRpV9ktw+vQ+uneX67F+HB2hbFkRV678bfzCdGPup0ePeL2YJUtCyZIiPW/b/5AxnD69n549lXpMopo1QSJREhj4MKnt8uW/KF9enJTkQad02b27nFOn9n7zcf6rE72VlRXh4rQbUOFikcGjzhmKmSUrEQa4ZhEikcmuN1ZZshIhShszXCjUiykWixGamRGduh+QJZNco743hEJJGo9RgPBw9OzspFIpSqUGuTxtP1M/S5nMkvh4jR5D5VPMLFlMV13NKKysshMerl9n1mohIkKLpaWJ3yWrbISH67dptRAerjb5Of2bYWmZhYgI/baEBIiJSdR7nlZWWQgPT1u5CAsTYGX15baMX4p/daL38GiOn5mAlOuYv4DDWnB39zQppqdnC/40MyNl9fg8cEILTZuatrvevGUbdgiF3EnRdho4r4UmTZomtZmbm+Ph6s44c3M+MYgVwHgLKa1THUX/WdCjx1CWLdMpIX6Cn59ObXFOCpG6HDmsqV69GlOnivlEm4+NhcmTZbRq1RtTkDt3bipW/IXp00VJdpcxMTB1qozWrfukf3EmoHXrHixZYsGrV8ltW7dCXJwVlSp9+altXcyeLFwoIyQkuW39egFgbVTj6T8YR5s2fZkxQ8b7jzxlrRbmzxdSrFhxPR2sihUrExdnxdYUZ7hevgQvLymtW5u23/Il+FfTK21s8uK1ZhON+vagrNAMLXBXo2HFus0m+z7my5efhSvXUW9AHyqIhCRo4T5aVm/YZpKhMkCBAgWZ67WC2sMG4CISoQQeAus270hTp52+0IvuL4Mp8iCACkIhFxMSqFGnHgOHjjTp3j8aAwcO5ciR/ZQpc4NKlSAyEp48gT59RqbZYF24cANdu7rj5PSCUqXMuHAhAVdXD7p1S2v/llEsXryZLl2asmfPK0qWNOP8+QQ8PFrQoUOXz1/8lahatTo9e46ldOnZVKtmztu3EBEhY8uW/SZruNSqVYeOHYfj7LyA6tXNCQ2FqCgrtm7d+9NaI/5ING/eioCAqxQtupkaNcx58kSLmVketm3boddPKBSyadN+Ond2Z9myeGxs4NIlFcOGjaVq1erffJw/1YpeoVCwbMkCGlQpj0uxYixZOJf4+PjPX5gOatasTY++A3gukxEks6Rnv0H8+qthDv2VK5fp1sqDmuWK06tdS27evG6wX5069enaqy9PpFJeWVnSe+AQqlX71WDfDRvWUrGYAzlFIqqUKMyePYYt6uo3aETn7r15bCElxNKK3oOGUrlyWhGqrFmz8efhk2w8cJSmXivwPX2RVVt2mMyR/oRz587QsWNjatYsTv/+7bl/P8Bgv+joKObMmUzduqWpUqUE69ev0jsdaQoOHjzDkSMXsbRsirNzR+7cecn48ZPT9MudOzd16rgTE6PhxIk4smTJSsOGLb5K2MrGxobDh6/g5eVLhw4bOXbsGnPnrjAYMzw8nGnTRlO7tjNNm1Zm27ZNRk8ZHz7sT+vWdalVqwTDh/fk2bOnBvv16zeMv/56SMuWqxg/fgd//fXQqIn4wYN+tGpVm1q1SjByZB+Cgl4Y7Dd48GguXrxP8+YrWbjQl4sX7xslCfj57aNFi5rUqlWC0aP78fJlsMF+Pwtevgxm9Oh+1KpVgpYta+Hnt89gP7VajYdHY4oXz4ajY1bKlbPn7NlTafoJBAI6d+5P3bqNuH5djLm5HcOGTTNI1S1evARDhkxELrfmzh1zPDxa06LF9znU9dOwbrRaLe3cGyG7fZNRCgVmwEILCyJKOvPnoZMm/TJrNBpaNamP9f17jFQo0ALzLSyIK1OOnQeO6q1wTp06weBuHZgRH09V4BwwVSplw869en+R1Wo1zRvWwS7wIcOUChKBeRZS1C4V2bb3gF7MefNmsn7RfBYBldCVY8YD0xbqq/4lJCTQ7LdaFHr6mCFKJSpgtoUU86rV2Ljz26/E/Pz2MnlyP+bMiad8eTh2TMCcOVJ27Tqqd0pRoVDQtGk1ypULpn9/FbGxMGOGlDx5fmPFim+v5zF8eH+OHvXBywtKlYIDB2D2bFi7dhcNGjT6fIDPID1Gx4cPMTRsWIn69cPo0SOB8HBd2ahs2VbMnr1cr+/GjatZu3Yqc+fKcXSE/fuFrFwp48CB8yafY1izZilbt85h9mw5xYrBn38KWb/eCn//i+nSatObk7f3Qv74YyGzZ8spXBh27RKyeXMWDh/+Czu71B5u/wykN5+QkNc0blyVbt0+0KqVmmfPYPx4GW3ajGTgQP033l9+KYxMFs6CBWBnB5s368xM9uw5pSdB/PJlMK6u1enZ8wMtW2p4/BjGjZPRtet4evfWP1Q3deoILl/ezvTpcmxsYPNmMYcO5eTIkcvpvu3/q+iV58+fZVLnttyOi0uqN6kBF0srxm3YkkZeNiM4deo4c3p04XpcLJ/O2iUCZS0tmbF1p97p2IbVXJjy5DEeKa7fAawqW469x88ltR09epilfbtzOS4u6XUpAXCWWbJo5149Kdjidtb4JCaSkm27GZhoacXN58lF1AMHfFk/uC8X4+L4lNJVQHGZjJV7DuDiUvGL555RaLVaqlZ1ZPPmt3ouS97ecORIXTZt2p/UtnPndvz8RnLsWFwSCyE+HgoXlvLHH6eNrkQzAwqFghIlbDh3TkuFCsntc+fCxo22XLz4JQrahpFeElmzxps7d2awa1fyG2Z0NBQuLOHkyZtJKzyFQkGFCoU5fz6W4sWTr580ScirV22ZP39V6tCfhVwup0KFwly5IqdIMtOWMWNEREZ2YtYs496pxuYUGxuLi0sRbt6MT/JtBRg+XIxS2Z2pU0038fmWSO8zmjp1FBLJBhYvTn7DfPECKlSQcv3606TN07/+ukjbto0JDoaUFdwOHeDKlQJcupQsqjZ+/GBy5vRh7tzkmE+eQOXKMm7ceJakhPrmTSi1apXh6VMlOXKkjGlB0aJjGTjQsB+AoTn9X9Mrb9y4RhOFQm9TQQi4xsVy/fo1E2NexzVFkgfdpkUThYIbN5JjajQabj15TNNU17sB1wP0Sxg3rl3BLUWSB52oWGOVSi9mTEwM0YmJpFaVcQfC4mL12m5c+Rv3FEkewBxopFbrxfwWiIqKJCIikl9TVZ7c3XXPLyVu3ryIu3ucHtVMKoXffhMYLXNlFu7evY1Eop/kQaeiGBb29pveG+DWrbO4u+uXEbNlg2rVzLl1K3lrPjg4iBw50EvyAO7uam7cuGjSvZ89e4qtrVAvyetiJnLz5gWTYj5+HEjBgmK9JK+LmcCNG+dNivmjcePGeTw89MuIBQuCg4OIJ08eJbX9/vt2ypTRT/KgU8+MiXml13bz5sU0MYsWhbx5hXrluDt3blGpkrlekgdwd1dw8+ZZ0yeVQfw0id7W1o77KXQzPuG+zNLk10hbWzvuy9IeELlvYYGtbbIVuJmZGTZZs/EwVb8HgF2q05G2+fITIE1LZXxgIcHOLjmmlZUV5uiExPTuDVgKhXptdvYOBKTUa/3UVyzG1vbbvkJbWlohEJjpMT9Ad8jGzk7f4crWtgABAWlP+N2/b6b3PL8FHBwKIpeThop5/z5Ipd/eICRv3kIEBOhzGzQaePBAo/e558qVi3fvEoiJSTtOO7tUDtsZRJ48NoSEqPRE2gAePABbW9NOQ+fNa0twsIrUW2ABAbrP+WeEra0DqdZlyOXw8qVKTyKkbNlyPH6s8zNIiXv3QCjU/z20tbXn/n39frGxEBqqIk8emxT98vHwoZrUWzYBAUJsbQ0732UmfppE37SpB7fNJawWCEhEV7ZZD/wtFuHhYRoV0sPDk8tiMes/xksEVgsE3DaX0LSph17fnv0G0Vsq4/XHn4OA/jIZPQfpm094erbgrFjElhQxvQUCHkgsaNTINamfmZkZFStVoRPwqUjzDOgJ1EtF7WzRsjVHhSK2ozMSSQCWmJkRJLPMlNpzejA3N6dTp2707Cnl3UdJzkePYNgwGT166FvFtW3bmT17ROzdq6OZqVQwe7aQ2Ngc1KxZ+5uO08bGBkfHInTuTBJPPCAABg+Gli37ftN7A3Ts2It168QcPaqbe3w8jB8vIkeOApQrl/yaYW2dk8aNG9O7t4TISF3bzZswaZKMbt2Mv76nhzx58lCnTn369pUQFaVru3ZNRwPt3t20mLa2dlSvXoP+/SVJf5QuX4ZZs6R07/7zGa4AdOs2jJkzpVy5ovs5OhoGDJBQMboDhwAAIABJREFUo0YtvYVIjx690WjMGDpUZ2Ki1cLp0zrpjX79xuvF7N59OFOmyLj28cU6Kgr69pVQr95veuKApUqVxs6uGKNGiZDLdTFPnIBVq8zp3LnfN5/7T5PoZTIZO/2OsLF4CfJKJOS1sGCVY3F2+h4x/SCSVRZ+9z3Maqfi5JVYkFciYWPxEuz0O6LnMgQwcNhIKnfrSSkLKYVklpSXyqjfZwDde+knkWzZsrNj3yGWFnXE1kIXc3vJUuw6cFRPyQ9gx96DCEqVpghgA5QC8lergfeq9Xr9rK1zsmPfQeYXLpIU889Spdnpd+Sr2TQZwdixMylUqB1OThYUKmRJ9epWdOgwLo06YN68tmzevJ9Jkwpga2tB3rwSjh2rwI4dh/UUB1NCq9WiSqlv+xXYu/csr18Xxt4ebGygUiWoUsWTSZOmGb0mMTExQ/r7nxtnkSLFWLXqdwYOtCVfPil2dhJu3arK5s1+aTbL585djUjkRqFCEgoUsMTVNRujRi2gTp16RqJ/HgsXrkOtbkzBghIcHGQ0a5aD8eOXUL16jc9fbAReXpuJjf0Ne3tz7O2ltGxpzdSp3lSuXMXkmJ+g1WpJSKlBnAnQaDTpxqxSpSpTp3rTooU19vZSHBzMiYtrwJIlm9L03bnzBH/8ISJXLsiZU1eqrF+/OX37DtTr9+uvNRk/fjEeHtmxt5dSoIA5anVjFixYp9dPIBCwYcM+Hjz4FTs7c/Lnl9G3b168vX2MMp4yEz8Vj97JqTj+Z/8mNDQEa2tLJJK0hsJfipIlnTl8/gohIa8RCARGSwwCgYB8BQuSxdKS4IhwHHLnIV/BQgYZL6VLl+HopWu8fv0KoVBo1D5QLBbj2a4TLxfM4U1UJAXy2NCiQyeDDKKyZctz/K8bvH79CrFYbFSN8ltALBYzY4YXY8bMICwsDDu7fEZFmD58iCE6+j3v3ikQCuHt21CD9EqNRsPKlYtZu3YpYWExODnZM3r0HJo0cTMQNWOwsrLC3b0tq1cvIywsFicnBzw9DUvV3r8fwNSpg7lw4ToSiYgWLZozZcrCNOcS1Go1S5fOZcOGlbx/H4uzcyHGjp1nULO/du26XLz4kODgICwtrYyexZDJZCxZspGpU6N4//49+fPbIzbk3PGFc1+2bCvR0bqY9vYOJtkC6kPAq1evUKlUhIRAzpxqwsLCP39ZOlAqlcyfP5lt2zbz4YMCF5cSTJy4xGS/WtBtcM+dO5EdO7bx4YOCihWdmTRpiUEr0Dx58mJjk5tbtx6RNasFdnYFMDdP+12uUOEX7t9/z7VrVwgKeoGbWzOjiyobGztsbHJz584TsmeXYWdX0GDfXLlysWWLHxEREcTGfsDe3uG7edr+NCv6lLC1tSN/fsOSsqbCzi5funVkn60bWT9lInsiwkkEtoe9w2vsCPb8ucvoNfny5U/XiX7j+tVsmzUVv6hI/tfeeYdFdXRx+F2WumAlookNKa5dYy/RmNijJsaOvUZU7Cb2Fo29I/bee+8Vo6jYe7w0FRQEFJTedvf7YxG4wCIsIOXb93l8opO5587s7D07d+bM78QB2wL8WTB+FCdOHE21vkQioVSp0l/VySfFzKwA5cpZaXTyL196MXBgZ8aNCyU8HPz8oHHjN7RrVz/FrHn58rmcObOYixc/ERurYvlybyZNGoSzc8pY5fSyaNFMLl9ezrVrYcTFwbJl3owf3x8XF/Hmob+/P127tqRLlzuEhirx8IhBIjnMwIEpk6LPmzcVFxdHrl0LJSZGxdy5Xowe3Ydbt26m2gaJRELZspbpOnBXqFBhypWzyrSTT81m5p08/PprQ2Syhzx7pl6Gc3KKYd68CRw/nvr3Mz1MmDCEly+38uBBJFFRKsaMeU7//r+LdGEyytixA/D13c6jR5FERqpwcHhKnz6/4enpLqr39OkTBg/uwsSJAtHRKu7ejUQQNjFlygiNtmvXrkunTl01OvlHjx5gb9+NadPciY5WcetWOI8fr2PGDM1LZubm5pQta/lVE5fnmfDK5Hxtlbp6VWzYExBA0jnCZWBkmbJcuftE02UaUalU1KpQjuPBQXyfpPwsMNnahnM372eyxV+ffv26Y2p6mv1JfvtUKihXDuztFzNwoFqKICYmhmrVynDnjjo++zN79sD69XXYv/9Shu8dGRlJjRqWPHwYSZkk+4/btsHu3Y3YuTMxD+7SpXMJDl7G2rWJIjYKBVhby9i8OfFsQHh4ON9/X45nz6JIsp/Khg1w9GhTtm49nuF25jY0PUcvXvxHq1b18PNTRw99ZsEC2LatFFevPk9xzZfw939HkyZV8faOJqmszpw5Ujw8urJo0boM23zzxocWLWrh7R1F0tXW6dOl+Pn1Zu7clQllo0f3p3r1I/z1V+Kk49MnsLQ04saN/7Q6DT9sWA8aNTrF6NGJPuzDB7CxMeL2bTetT8Mn5f8qvDInUSqVvAwIIHmitXqAp+/b1C75ItHR0fh/+kjyFBL1Ac83PlrZzGl8fF7QpIm4TCKBBg3USZw/ExwchL6+UuTkAerVQ+MJ0S/x/n0gMplE5OQB6tcHT09xbJOX1zPq1xcrlUmlUKeOFA+PxFmgv78fRYpIRU7+s00vL/FsMb/x4MFdypQRO3lQj2VwcCoqc+ng1atXlC9vRHLttPr1FXh5ZfyHQ23zJZUqGZJcs09tUxxi4+n5nPr1xW+WhQpBuXJG+Pi81ur+Xl4vqF9fPFE1N4dSpQzx8ck9z7HO0acDPT095N+VJHm0qzNQ0VK70CgjIyNKf1OM5AsAV4AKVtapXZLrsbKqLsoHCuqZsrMz1K/fMKGsaFFzwCBFWJqzs/qYuDZYWBQnOlpPJH4G6tylFSqID2pVqFCLK1fEG+MxMeDiEkfFipUTyr79tiQfP6p49Ups88oVCRUqZDxXQV6ifv1GvH6dMlz18mX45hvtQmWtrW0QhOiEaKPPXLmiT4UKNVO/6AvY2pbn6dPoFOGqV64YpLBZseL3XLkiDgoIDISXL2Ow1PI5rlChBleuiN2onx+8eROLpaWlVjazA+nMmTNz6t6FgdGRkTFos3pkampERETWRGukh8LFLBh/+QIV4uIoBlwAhhibMHXRco1pw9JCIpFQoIg5f169TKW4OL4BzgFDjU2YucxRlEwkr1C7dl2mT19NWJha49zPD+zt4c0bM1at2pqwcS2VStHXN2D69FtUrRpLoUJw9CiMG2fC3Llrtdp/0dfXR6VSMWvWHapWjaVgQTh0CCZMMGH+/LWisxY2NuWZOXML4eHRVKigwscH7O2NKVGiEQMHJkZVGBgYoFDEMGfOfapVi8XMDA4cgMmTTVi4cF2a+y95BU3PUZEiRbhw4Tg7dwZSsyYYG6uVM2fMUIvHafP9lMlkvHv3BicngerV4zA0VMsKLF1qwpIl6ymS/DRROjAzM8PHx4t16zyoXj0OfX310pqjo4ylSzdQKMkrSblycsaM2UGhQrFYWanDb/v1k9GyZW9at/4tjbtoxtKyPKNG7aJo0VgsLeHJE+jbV0b79oNo3vwXrWwmJ/kYSSQSZDJDgBXAx/TY0K3RZ4ATJ47iOPdvPHy8qWBZjlHT/qZVqzap1lUqlTx79hSpVErFipU06tEcPXKQVfPn4OX7lorlrBgz/e9UIzryCvfu3WHkyF74+PghlUKFClXYvv1YCqVJlUrFnj3bcXKax5s3AVSvXpEJE+ZnKhxQpVKxc+cWVq9egJ/fe6pXr8ikSQtFbxOfefXqJXPnTuDixSvIZEb06NGPsWOnpgiBValUbN26gfXrF+HvH8z331dmwoSFWssEa0NMTAzPnz+lYMGCWFnZpFnXx8ebwMAA5PKK6coWldZzFBcXR//+Xbl16zKRkUosLAoyZcoSOnXSPum2QqFg9epl7NixhqCgEBo2rMtff83XmM0tMjKS//57hrn5N5Qta6mxnU5OS9i5cx3BwWE0alSPiRMXpCq58fjxQxYsmMStW3cpVqwwffoMx95+pMaNUS8vTz59+kilSlU0BiE8eHCPefMmcOfOA4oVK8ygQWMYNGhYlm225gqtG7lc/gq17HlUfNEEQRDOabwgEUvymKNPL7du3WTUoD7oh4cTp1JhVLQoq7buSjOpc27uj7ak1aeIiAjGjRvIlSsXsbQ0wssrhsGDhzN+/HStRdrCwsIYM6Y/1645Y2lpxMuXsQwdOppRoyamsHnmzCkmThxK0aIKPn5UULx4Gdau3ZfiFT40NITRo/tx48Z1LC0N8fKKxcFhPCNGiA+LZRdHjhxg2rTRFC8O79/HUbasLWvW7E2hjhgcHMTIkb25f/8OpUoZ8upVHH/+OY1Bg4anaT83f+927tzMP/9MoVQpPfz84qhcuRqrVu1OM/9vVvXHz8+XYcO64+EhULy4Pn5+MGvWEjp37p6i7saNq1i0aA6Wlvq8eRNDzZp1cHTcmSUbsZC7NmM7C4JQI/5Pepx8vuXDhw/0t+uIY4A/QngYnhHhzHjjQ8+O7QkPD8/p5uUapk0bhb7+Rby9o7l3L4Rnz6I4d24Ne/dqr3I5efJwChRwxsdHbfPRo0iOHl3BoUPiEFh3dzfGjRvAoUMfefo0lNevI+jZ043evdulCAP9668hfPPNv/j4RHH3bggPHkSyf/8SjfK2Wcnjxw+ZNs2BM2dCefw4FG/vSNq2fUb//r+RfII2alRfrK1d8fGJ4v79EG7fjmDDhtlcvJg3H0cXl2ssXTqZa9fCefgwlDdvIqlV6z5Dh2r/NpFeVCoVAwb8TrNmT3jzJpJHj0K5cCGUv/8elUKz6cKFs2zYMIfbtyO4fz8EH58orK1dGTUq+/MVZATdZmwWc+TIAdooFLQFJPF/ugO1FXGcOpX3w/GygoiICI4ePcKqVdF8Xl347jtYuDCC7ds1Ky2mRWhoCKdPn8bRMTohAqN0aZg/P6XNPXs2MXhwLA3iz+hIpTBmjBJj42Bu3kwUFgsODuLChQssXx7DZ6mhsmVh7twItm9frlU7M8KuXesZOTI6QahNXx+mTFEQFuYrimJ6986P27dvsWRJDJ9XF2xsYObMCLZvX5mK5dzP9u2OTJoUQcX4vXkDA5g7NxZBeKZ1ZFZ6efbsKR8+vGbGDAWfjyNUrw5jxkSxY4dYXXTHDkdmzYrAJn5FzcgIFi+OwdX1Ju/e+WVrOzNCVp2M3SWXyyXAdWCyIAjp2iAAPr+CaEWxYtpJH2Qn4aHBWEdFpSi3jokhKio0zTbnxv5kltT69O5dOIaGeiTTg8PKCj58CNLqc4iMDMbUVErhwiltBgV9ENkMCfGnRg3xaV2JBKytJURHJ45RSEgAhQsbULBgTAqbHz68z/bxCgnxx9pa/IYhkYCVlZTY2PCE+/v6elGihCHGxuKQUSsrdSjkl9qZG793nz4FplDjNDCAMmUMUCgisvU5io0No1w5A5IvsVtbq3Bx8RfZDwoKSBEmbGICJUoYolRGZtlnm1k7WeHoGwuC4COXy42A5cAqoFd6L85va/TVatRltqkpU8LD+XzeMQo4ZmDA+qq1NLY5t/YnM2jqk0RiQpEiRbhyxY+ff04s37dPj3r1ftDqczAyKoShoSnXr0eIJJX37dOjbt3GIpvff/8z+/adZcCAiARJ5Q8f4MqVGKZPr55QVyYrilJphKtrOPWS7L3u3Sulbt0fs3281O28TrdukQntfPcObt+OxsqqUsL9zc1LEhioVlesUiVp3w2oXfunNNuZW793NWv+xL59T2jVKvFH1sMDPDxi+fbbctn6HFlaynnwIJo3byBpANi+fcbUrNlcZL9OnZ/Yt8+LRo0SNXaePFEL66nHJfOfbRpr9Okm0+GVI0aMCIn/r2LVqlXewIQRI0ak5/07T4VXppeyZS258K8zO94HUjQujmfAEJmMCk2bMWiog8brcmt/MoOmPkkkEkqWtGbIkFMYGCiIioLVqw3YsMGUFSu2UKRI0QzfS61TVJYhQ85iaKggMhIcHQ3Yts2M5cu3iDbGbG3Ls2XLUc6fD6Nw4Thu34aBA2V06DCQtm0TlUP19PQoXrwU9vYXMDKKJTwcVqwwYM+eAixfvlkUupcdyOUVWbXqAC4uERQsqODGDRgwQEbv3iNp1ixRtVRfX58iRcwZPvwqMplaAnnRIkNOnCjEsmWbEhJqpEZu/d5VrFiFuXP38ORJDKamCpyd1WM0atQM6tfXnGM1K/pjYmKCSqXizz/vUqhQLIGBMHOmEQ8flmDBAkdR9E3FilWZNm0X7u5xmJgoOHcO7O1lTJy4gBo1sibZeo6HV8rlclNAXxCET/FLN3OASoIgpEc32JJ8GnUTExPDrl3bOL1nF3pSKb/17ke3bj1SVXBUKpW4uFwjMPAtpUpZU6dOXY1RJ/7+/ly+fAF9fX1atGiVZbv62cWXxujevTssWDAFP7/XVKlSl8mT/0kz7V16uHv3Nps3L+ft25fUqNGIP/4YnWr+zrCwMLZuXc/ly4cxNS1A165DaNfut1Q/e1fXW2zZsozAwDdUq6a2qUkX6datG+zcuY1ChQoxevSfGiNEYmNjuXLlIu/evaNWrTpUrlwl1XohIZ/YvHktV68ep2DBItjZDaN169Tjs52dL7N48Uzev/enYcOfmThxlkgqNyleXp64uFyjTJlvqVu3CSap5DvIKJ6e7ty44ULhwkVo0aJVilDVjPLhwwccHAbh6noDU1NT5sxZ8kVJ8vT4hbNnT3P8+BFKlizFqFHjNP4Qnjp1nJUr5xIS8pHmzdszfvwUChUqnKJeQEAA8+dP5+5dZ4oWtWDMmFn8+ONP6e/oF8jx8Eq5XG4FHEKd7EmKOm/GSEEQ0rMLYUk+dfTp5cOHD/Ts2YbY2DfUq6fg6lU9vv22Elu3Hkvx5du+fSNz5kyiZUspMTESrl5VsmzZxkypPWY3aY2Rn58v3bq1wszsA9WqKbh0SULlyvVYt26/xnjlnOZL37mOHZvx4MEd2rYFf3+4fRtmzFjCgAGDRfW8vDyxs2vDd9+FU758HOfOwQ8/tGTFiq0a5Zy/hIeHO3Z2bShdOgJbWwVnz8JPP7VlyZINIpsqlYrZsyewb99W2rTRw89PypMnemzffkykm58RVCoV06eP5ciRXbRpI8HHR8qLFwbs2HFclFM4IygUCmrVKkNYWCht24Knp3ppauLEuSmkgpPypXMBzZvXxNf3Fb/8os6r8OKFhNWr96T48Xzw4B59+3agShUF332n5MwZJd2792fq1PmiyYBCoWDcuMFcvnyKNm3A3V2Kj4+MvXvPYG1tq1Xfv9SnfJ0zNjn5wdE7OPSmePHTrFgRi0Silgvo1csIc/P+zJixMKGeh4c7v/7aCFfXqISNn3v3oEULE27efB4vKZD7SGuMevduS716N5g1SwFAbCx06GBCjRpjGT16wtdsZrpJqz9r1qxizZrJPH4MnyfxZ89Cp04Snj59K/rhbt++Id27P0sQwoqMhJYtZbRvP5v+/QenZv6LtG5dh4ED3Rg2TG0zIgKaNZPRrdtCevbsk1Dv3Lkz/PNPf27ciEhIa3foEIwbZ8GtW4JWPzQnTx5n6dIhXL8enqCNs2cPTJ/+HdevP9fq4FC3br/j63uJO3egQPw+5LZtMGIEuLkFa2xnWmM0duxIHj7cyq1bJER7rV8Pkybp8/z5+4R2KhQK6tUrz7JlgXTsqK4XHAwNG8qYOnULLVsmHpLcuXMr+/dP5NKliIRoLycnCZs3yzl79naG+52ePulEzfIQcXFxnDhxmpkzYxM22qRSmDUrmsOH94jqHjmynz594kS7+7VqQfPmepw9e/ortjpr+PTpIy4ut5gwQZFQZmAA06ZFcuTIthxsmfbs27eWv/5KdPIArVuDjY2KzZvXJ5S9eeODl5cHDg6JkxsTE5g8OYIjRzZrdW8vL0/8/LwZMiTRpkwGEydGcOTIJlHdw4e3MG5chCh3aadOULBghNa5l48c2cL48eEiAbTu3UEqDeHx44da2XzyxJkZMxKdPECfPmoHvWaNo1Y2nZ0PMnNmopMHGDQI9PTiuHjxfELZ3bt3KFw4KsHJAxQpAmPHRnD48FaRzSNHNjN5coRIVM3eXoWv72tevvTSqp3Zgc7R5xAqlYq4OAXJlzFlMoiOFof+xcREI5MpSI5MpiQqlVDO3E5cnAI9PQnJJb5NTSE6OvdtDKaHuLgYUlMckMkgPDwi4d8xMdEYGuqRfEIqk6n/nzbExMRgbKyXIhxQ/XlGJ6sblULpUX1/SSbuH5nCpkSithkVpZ1NUKVq09gYwsK0e5NXKhUpbOrpqW0mPcyoft5S7tWox0j8vEVHR2uwKUnx2eckOkefQxgYGNC0aQPWrhV/oRwdpbRqJc4D26pVO7ZvN+HTp8Syt2/hxAlVtueMzQ7Mzc2Ry23YsSOxTKUCR0dDWrXqkHMNywTNmnVmxQpI+mw/fQoPH0L//gMTysqVs8bMrCjHk5ydU6lg1SpjWrbsotW91anozDh7NrFMqfxss6uobqtW3Vi9WkbSjHt37oCnp4ratZMLcaePli3VNpMmErtxA3x9JdSsqV3kybffVmbZMvVy5meuXoWAAHBw0C5nbfXqTVm6FFGC7osX4dMnCW3bJu511alTD3d3RUIeWFAvLa5ZI0vl8+yKo6OxyOaZMyCVFvoqKQLTi26NPgfx9HSnU6fmNG4cTf36EVy+bMqzZwU5etQ5hTLitGljOHduD/37RxIdrcfmzYbY209g2DDtkj9/DdIaoydPHmFn15bWrWOoUSOK06dN8fW14MiRK1qFV34N0upPTEwMjRtXRCIJZPBgdbz7xo3QtetAFixYJqp769ZN+vfvSMeOcVSoEM2RI6ZERVmyf/+FNEMh08LF5RqDBnWhUycF5ctHc/iwKUqlDXv3nhPlP46NjWXAgI74+9/Fzi4cX19DduyQsmTJBtq2/VWre8fExNC3768EBz+ie/dwvL0N2b1byooVWzWK/n2JwMBAGjWypVQpJX37qjdjt22D9u3tWLVKc4KStMYoODiIxo0rUaJEBH36wIsXsGsXjB8/ixEjxD8ep04dZ9y4wfTqpaBkyRh27zalRInabN58WJQRLDw8HDu71ujpedCxYzhubkYcOiRl06aDNGz4Q/ImaMX/5Wasq+stTh7ej7GxAT+3/pX69RtqLYKVXSiVSjZsWMPBg1vR05PSu7c9vXr1S7Xuy5cvmTlzEn5+7lhZVWPWrHmppgpUKwn2wtX1IhKJHm3bdmPpUu3WKrMTpVLJihVLOHVqLyYmpvTvP4KOHVOfqb548YK//56Ev/8rKleuz8yZc3LtxjJ8eXIRFxeHo+Myzpw5iExWkNGjJ9G06c+p1vX3f8e+fbvw9/ehTp3G/PJL+0wner9//y7z5s0gKMiPunV/ZsqUman+cERHR7No0TyuXDlBkSLfMGrUJBo3bpqpe6vzts7h+vUzFChgzrhxU/nhhyZfvjANwsLCGD78Dx4+vI6JSUGmTZsnmnknZc+enSxe/DcKRTiVKzdk3botqfY9JCSEIUMG8t9/NzE2LsyMGfNp06ZdqjZ3797B0qWziYmJoGbNH3FyWp+qIuiHDx+wtx+Am9tdTEwKM2PGAo02teH/ztHP/3saBzdtYHBUJFJgvbEJ7Xr1Yfo/C7947dekVau6+Pu/wMFB/crn6AiVKtXn4MHzonpubgKdOzenRYto6teP4vJlGXfvmnH0qLMo9luhUFCzZhn09EJxcFAvDzg6grl5Ga5ff/q1u6cRpVJJkyaViYl5y7BhEBICq1ZB48Zt2LRpn6juw4f36dmzHR06xFK9ejRnzsjw8DDnyJGrWqV0+xrk5rfIq1evYG9vR48ecZQvH8PBg6aEhZXlwIGLIof3efYdEvKInj3D8fXVZ/NmA+bPX8Ovv3ZM4w6aiY6OpmfPtsTEPMPOLhxvbwO2btVn8WLt3xIyQt++dvz77yn++AMsLdUa969f6+Hi4i46xxAREUG3bi0xNPSkS5dwPD3VbzOrV+/k559biGxOnfoXu3atxd5erZm0cSMEBRXg+nVB9Hn6+/vTtGkVbG2j6dkTnj9Xv3mMHz8bB4dRWdK//ytHLwgv6NSiCc+jovg85wsGKpuYsOvURY161l+bvXt3MXv2UNzdE9OwBQSoRaY2bToqmuF1796S335zZeTIxP5PmSLF07M9jo7bE8pmzJjC4cOOuLsnRiG8e6e2uXz5Vn77TbsHNKtZvnwJu3fP4tkzEjaovL2hQgU4ffqG6FBQ69Z1GDNGoFcSsYxhwwxQKPowZ84yciO51dErlUrq1bNlw4ZAWsT7K5UKunQxpkKF8Ywa9VdC3V27tnP06F9cvBiRINj14AG0aGHK/fteWh2c2rp1I+fPT+Xs2YiETebbt6FduwLcv++VreciAgMDqV3bmqtXoU4ddZlCAT/9BKGhNTh//t+Euk5Oy3jwYD5HjyZKSjg7Q+/eRblzxzMhZDM4OIjq1S25eRNqxCuLx8XBDz9AuXI9WbkyUdise/f2mJld5ehREmyePw+dO+shCO+zJEn7/1V45YUL5+iqUJD0xb4IYBcTw4ULZzVd9tXZu3cH/fuLc21aWECXLuoH4jMxMTFcu3abP/4Q/8gNG6bg/HnxzP/48f0MHiwONStRAn7/HRwdc49TPHlyL/b2iKIQypSB5s1h06bEEMP379/j6fkSOzvx9cOGxXLx4omv1Nr8g5ubgFQaQfPmiWUSCQwdGsXFi4dEdS9ePMiQIYlOHuD778HWVo+7d7WL+7548QD29hGiSKK6ddVj/+BB9ia5X7FiCWXLJjp5UIcpjxwJb98+EdW9dOkQQ4cmOnmApk3B1DSG588T88vu2bOT8uUTnTyolUNHjgRX11Mim8+f32TUKEQ2W7QAMzMlFy7kHonoPOPoTUyMCUnlkESovj7Gxpk/vp1VGBuLo2NtLtxKAAAgAElEQVQ+8/EjyGSJ63tSqRQDAynJJepDQ8HERLxWa2BgpNGmiUkqsXI5hKGhcartDAkBMzPTJPUMUChUJI8MDQ0l08fm/x8xNjYmIkJJMin9+M9TlqyujNBkLyUqFYSGqrT+7I2NZSlytn62aWKSveNZqFAhwsJIoZcVGgoSidi9pdZ3pRLCwhSiNxkzswKEhaW8V2goSKXiZ1Mq1U/VZkQEFCxYMMP9yS7yjKNv164DJ4Ckxy+eAockki/qX3xNRo/+kx074L//Esvu3YPTp2Hs2MRXaKlUym+/tWfqVMOEL2lcHEybZkSnTj1FNseMmcjmzeoj259xdVWHhs2duyg7u5MhhgwZw+rViJJpX72qbuuIEYnRQQULFqJx40b8849+Qt+jo2HmTGM6duz3VducH7C0LEepUuVYtSrxcQ4Lg7lzZXTqNEhUt1On/ixeLOPDh8SyffsgMtKMWrXqoA2dOg1i0SKZKOn39u0AhdPMqpYVjBo1jo8fEYXqBgfD339D3bpiWYOOHQcyb574R2ntWgnFipXC2joxRWOPHr0JDNRjX5JtpffvYc4cUnyeTZt2Zto0RD8MK1eCoaFRptJiZjV5Zo0e1DlbxzsMoZFUih7wb1wcC5Y78buGqI6cYvx4B/bv386PP6o3Y2/cUDvBKVNmieoFBwfRu3c7QkJeUrcuXL2qwtb2ezZuPCQKiQP47Te1jsqPP6qd4q1b0KZNZ9at0+40ZXYxeHBPzp07QbNm6pn83bvw558pN6b8/d/Rs+cvwDuqVVNx+bKSOnV+xMlpZ6ajT7KL3LpGD+rTsT17/oK5eRg2NkouXlTwyy8dmT9/tUiCQKVSMX/+NLZtW0fLlvr4+urh6WnA9u3HqVq1mlb3VqlUzJkzkd27t9CihRQfHz1evzZi584TVKpUOau6qJG1a1cxf/5k5HL1Zuz582Bu/g2uru4pdH4mTx7B8eP7adFCipeXHu/eydi9+zQ2NmJdmsOHDzB27CAqVlRRujRcuADVq9fi8OFLos8zLi6Oli1r4e39khYt1CGbPj56bNlyjMaNf8yS/v1fbcaCetd86JD+3L18EQnwfdOfWaMhjOrZs6esW7YI92dPsa1chSFj/tSoEJheDh8+wNKlU/n06T1FihRnwoQFqYZ7RUdHs2TJfA4c2IKenoTevYfh4DAm1Y0ZlUqFq+st3r9/S6lSVmkKS929e5t582ZjaGjEnDnzMi2aFBERwebNazl//gCGhoZ06DAAO7teqeqI3LnjysaNS/H29qBKlToMHTpeY6Lq58+fs2XLeszNCzNw4HCNCo5KpZIbN67j4+NNtWo1NI5PWFgoGzeu5uLFw5iYyOjceTBdunTPVPLlkJBPbNjgxJUrx5DJTOnadQidOnVNNVT3+vV/2bJlGQEBb6lWrQH29uNSVdn89Okj69evwtn5OAUKFKRbN3s6dOj01cJ/4+Li+PffKwQEBFCnTt00vx8+Pt64uFzD0rIkNWs2zJIf19evX3HzpgtFixaladNmGm0uWDCHAwfWEh0dSZkylVm0aC2VKqVM5J0RgoKCmDhxLEFB7+nZsx+//9451Xr+/u+YP386Li5XMDcvxpgx00TaNUkJCwtj5cqlBAYGYGfXi7p162u8/6VLFzh58hilS5dh2LCRWboE+X/l6JVKJXUrWlEqOIgpqNec5gJeBQtxz+216KF3db1Fv66/8Vd0NI2VSq7r6bHQ2Jgt+45Sr57mwUqL9evXMH/+BGbOhIYN1bv1c+bAP/+sEolGKZVK7OxaY2T0iNGjI4mLg4ULZRQp0oT16/dpfOi/9mwxNjaWLl2a8c03AiNGRBIRAfPmyShXrg0rVmwR1T137gzjxvVj2rQoatVSce6clNWrTTh06BIVKlTUeI+s6FNUVBQdOjTB2voV9vZRfPoEc+bIqFmzC3PnaneOICIigl9//YFKlXz4449ogoJg9mwZDRv2YNaspaK6Bw/uYc6c0cyaFUmlSnDsmD7bt5ty4sQ1ypa1TKgXFhZG+/Y/8P33bxk4MJrAQLXNn37qz9Sp8zLzEWQrX/t7169fJx4+vMCCBeq0jDt3wu7dEs6evZ0lJ0nT6o+/vz+//NKADh0+0bVrLF5eMGOGjCFDZjBw4NBM3zu7+L9y9Js2rcdp0ng8gc/zhFhADvSeLl4a+L15Y4Y+fkTSle5dwJpq1Tly8ZpWja1SpRiOjtF0TXICeutWmDLFlIcPE1WZL106z7x5fbl3LzwhsiE6GipWlOHkdFzjMfOv/cCdOHGUjRuHcv16eIJGSkQE2NiYsGfP5YRXbpVKRePGFXFy8hVFdSxZIuHatZZs2HBA4z2yok979+7i2LHxXLgQnhDZEBIC1tbGnDp1i3LlrNI2kAo7dmzhwoVJnD6dmGHq40ewtjbi4sX7lCpVGlDPkGvXtuL48Y/USnKSf+pUKW/edGPRorUJZZs2rePmzemi0L0PH8DGxoh//32c4qRzbuFrfu/evfOjbl05//2nXmL5TP/+4OZWl6NHL2b6Hmn1Z/bsySgU63ByStR/cHeH+vVl3L/vlWK5NLfwfxVeee7cabqS6OQBDFAn3k6e6d716RM6Jbu+E3DryWOt7h0XF0dAQLRIzQ6gY0cICBCHzdy+fZPffw8Xha8ZGUG7drHcvu2q1f2zA1fXq3TsGC4SwpLJoE0buH37VkLZx4/BvHv3nmbNxNd37qwS1csubt++TKdO4aLwtYIF4eefpVqHA96+fYnOnSNENgsXhh9/NBDZ9PZ+jb5+jMjJA3TurOD27X9FZXfvXqJzZ3Honrk5NGxoqLUqZH7j3LkzWFmJnTyolS5fv36S6jVZyZ07l+ncOVZUZmsLpUpJcXN7ke33z0nyjKP/9tuSPEul/BmkmC0VL1gIj2T1PIDiWqZ+09PTw8RErbchsulBCpU7C4sSuLunDPf08DDUmO0nJ7CwKIm7e8qDLB4eUiwsiif8WyYzRSLR49275PXAwiL75QosLErh7m6QotzDQyJqZ0YoVqwU7u7i/RKVSt2npPITRYsW5ePHuBQho+q+i8dSbVO8t6FSgaenMleNe05iZWXN27eIBNVAPas2Mcn+UEQLi+/wSOYYoqLA1zeWYsXy9xjlGUc/adJUrqFeglHF/9kPXACmTJkuqtvvj2EMMzEhMP7fgcBwExP6/TFMq3vr6elRu3Y9BgwgISzN3x/++APq1xdrmfz+e2fOn5dy+LD6QVcq1Us8Dx8a5KpsUF279uTQISmnT6vbqVDAmjUSXr40oXnzlgn1jIyM6NbNDnv7xBj5N29g7FgZ/fpppyKYEbp378f27QZcvqz+d1wcLF2qR2hoIa21VHr2HMjGjQZcvZpoc8ECPeLivqFevQYJ9QoXLkKbNm1wcDBKCJ/z8IBJk2T07Ts2mc0/WLPGEBcX9b9jY2HOHClGRt9qrQqZ32jc+EfMzMwYO5aEMxSPH8OMGTBw4J/Zfv++fUfx998mPIufMUZGwrhxhtSuXTfVdJP5iTzj6EuU+JalazcxQqqPBWABDJFKWbByDaVLlxXVdRgznqp2vShvZEy1AgWwNTKicveejBij/Zdp164TREaWp1QpsLZWv35KpVXYulW8Rl20qDnbtx9l0qSSlCsno0wZE5YssWL37tO5ag2wRIlv2bTpICNGlMDa2pTSpU3YsMGWvXvPpIiWmDZtIQUL/oqlpRFVqxagWjUTfvnFgR49+miwnnVYWpbDyWkn/ft/g42NKaVKmbBvX0V27Tqtddo9a2tbHB2307u3Oba2ppQsaczx41XYufNkikieefPWEBHRjDJljKhatSD16sno23cS7dqJNVzk8gosXbqFbt2KUL68GaVKGXP+fDW2bTue60T3cpKDB//l2LFCfPONejO2QQNo27YPAwf+ke33btz4R8aNm0/TpmZUqmRGqVJGvHzZkJUrd3z54jxOntmM/YxSqeTy5YsUKGBEnTqN0wyx+/gxGB8fb0qXLqMxkXZAQAD//DOBU6dOIZFAu3btmTJlgUZhrZcvvXjw4B516tRN8QOTFJVKFX80XYq1tY3Gh33//t04Of2Dp6cvFStaMnbsHNq0aZui3sOHDxg+vBuvXr1DKgUbm3Js23Y81Tbs2bOd1avn8fLlO6pUsWbs2NmphpDduePKyJE98fYOQCoFW1sbtm8/RsmSpUX1FAoF7do1w83tPhERaimG5s27sXr1Bo39h7Q3xpRKJcOG9efSpWOEhSkxNzfijz8mMXJk6rLLCoUCQXiBTCbD0rJcmvdND15eHsyePY6LF68hkxliZ9eLv/76W+OPsb+/P3FxYRQp8m2aP9hxcXEIwgsKFChAmTKavx/p5cWL/5g790+cnW9QqJAMO7t+jBs3LVP6Mc+fP2Pu3PH8+68rRYqYYWfXn7Fjp6T4gY+JiaFPn47cvn2NyEgVxYubMmHCQuzsemt977i4OFauXMDmzasJCgqnYcPvmTZtKdWrf6+1zaSkZ3M5OjoaNzeBb775RmOS99zE/9Vm7Gf09PRo3rwl7dq1+2IcdeHCRahatbpGJx8TE0Pnzs0oUeIYghDFf/9FUbToUbp0aU5cXFyq15QrZ0XHjl3SdPIAEokEubwCNja2Gp387t1bWb58LGvW+PDxo4J58zyZNGkA586dEdX78OE9nTr9RMeO73jzRr2mWafOS1q1qp2indu2bWD16r/YsOEtHz8qmDXLjT//7MflyxdE9fz9/enWrSV2dgH4+oIgQI0aHrRsWRtlsrP0rVs3IDj4PmfPqo+Bb9sGFy7sY+zYkWl+BmnRr18nnj49wtmzSkJCYM2aaFaunMnGjWtTrS+VSqlUqXKWOPn379/z++8/06SJM35+cdy5E0Fg4DaGDOmq8ZrixYtTo0aNL76V6evrU7lylSxx8n5+vnTu3JzWra/x7l0c//4bgqfnekaN6qu1TR8fb7p0aUH79i74+8dx+fJH/vtvLePHD0pRt337hkRE/IuLi4qPH2HOnHAmTRrO+fPaa0tNmTKS27dXculSKEFBSnr3vkf37m3w9HTX2mZGMTIyomrVannCyWcVeW5G/5msCAs7fvwIO3YMx9k58fyySgWNGpkxZMh6fvkl6zSlk6NSqahb14Z9+wKpVy9pm2DOnAqcOpUY/TFq1HD8/Xdw6VLS60Euh+7dE5MmKJXK+HDAIL5PMkE6dAiWLq3G0aPXE8qGDRtIePgBTidJOatUgpUVDBmykEGD7AGIjIykYsXiuLpClSTnmfbsAQcHPV68+Kixj5rGKCIigkqVSnD/vroPn9m2DaZOLcD9+2812swKVq5czNu3C9m8OVFsJzYWypUzYdeuyxpPc37tENj582cQE+OEo2NiesWoKChTxpiTJ7ULLf3774kYGm5gyZLEHdGICLXN8+fvJhwE8/T04Oefa+Ljo44e+szs2bB3ryXOzhmPYAsICOCHHyrz6lW0SPRv+nQpvr69mDcv8/kVcvPpZW35v5zRZyWC8IIffhCrF0kk0LhxZLaHW0VHR+Pr+0Hk5AGaNAFBeJmsnQ9p2VJcTyKBn3+Ge/fuJJSFh4cRHBwqcvIAjRvDixficAN390cJkraf0dNTy7smDQP9779nqFRiJ//ZZlRUMhWtdOLl5YG+vtjJf7YZEpKKmlQW4+Z2nyZNxIpqBgbQoIE+gvCfhqu+Pm5uD2jSRJxD19gYatUyxN1d0Mqmu/tDmjQRh73IZFCjhiEeHomz6tu3b1KmjNjJA/z4IwQF+aENL196IZcbkTz4rUkTBW5u2iUR15E+/q8dvY2NLTduiOUTVCpwcTHBxqZ8tt7byMiIb78tKspLCeDiAra24td+W9uqotn853Y6O0ONGolB3qamZhQqZMqjRyltli8vnv2VK1clVZtXr0KtWrUTysqXrwCoEyokt2lsrN0mo6WlFbGx6iWopNy4AQULppJhO4uxsamOi4t4jTsuDlxdFdja5p48n9bW1XBxEa+bR0fD/fsxKbRZ0ouNTTVcXMThqpGR8OhRjEjYq06d+nh7IxIqA7h2DYoU0S6s1dKyHIIQnULp8vp1KTY2uSOfRH7l/9rR//JLe968KcTEifoEBalDJ//8U58PH8y1znWZXiQSCSNGTKZXLxk3b6rDGy9dgqFDTXBwmCGqO336bG7elDBjBnz6pE5kMnQoBAYaik4E6+npMWLERHr0kHH7ttrm+fPg4GCCg8M0kc2ZM//B2VnC7Nnqk6bv3sGgQRASYsTgwYnHwc3MzLCysub339UCZQoFnD2rvn+LFskE5dOJmZkZjRo14bff4P59tc2TJ8HBAQYNmqiVzYzQs+cATpwwYvlyCeHh4OMDffsaIZd/n2sS2AD062fP7t2GrFmjdsavXkGPHsbUr99Yo87Ql20OY8sWQzZsUC8DeXlB9+7G/PhjM9G+go2NLba21vz2m1qoKyZGnV917lz46y/tJB2KFy9O+/Yd6NrVBHd39Y/W9u2wapURgwaN1sqmjvSRrx19WFgogvCCsLDU1+yMjIw4dOgy7u6tKFFCj+++k/L6dRsOHrwkSgCclNDQkHibmV9i6NNnIEOGzKNHDwv09cHBoRQzZqxJIZRWrFgx9u07z44d32BuDqVKwdWrpTl16lYKobQBA4bSv/8cune3wMAAxo2zZM6c9Smibr77riS7dp1i8+aiFC2qThJx61ZZzp69k2KT++LFu5iYVOann9QJGLp3h0aNfsPRMfWNU5VKhZeXBz4+Phr7vnPncaysWtO0qQR9fejf34DBgycybFjqG7xKpRJPT3d8fb+8fh8cHIQgvCAqueB9PMWKFePQoUucPfsDhQvrUbWqMTKZHevXa5ZzyA48PNw5f/4sIcmnuPGULFmKffvOceBAXUxNoWpVY0qU6IOj406t71m2rCV7955h9+46mJpKqFvXlLJl+7Ns2ZYUdU+evIlKVZeaNdWnu8eNM2LGjGUa87amh/nzV1O58hAaNDDFxATWravOjh3HsbXN/Bu0v78/L1680BhIkZ2EhYVx/vxZ3Ny0W1LLbjKd50oul5cHtgHmwAegjyAIX28LPRWUSiXz5k1h27ZNWFjoExAQR58+A5k8+Z8UTuzGjes4O5/D1FSJSgVXrpyldesu/PprB1E9hULBP/9MZMeOrRQvbkBAQBwDBgzhr79maa2iqFAocHN7THDwJ+TyAvj7f0AQnqBS/Z4iUqdOnXq4unolRMRouqdEIqFv30H07TsIlUqVZgx3w4Y/cPv2qy/alEqlXLhwM6HNacWv37zpwvjxA4mM/EhUlBJb2/KsWLEjRbSMnp4eW7bsB9TjldZneO3aVf766w9iYj4RGamkQoWKrFixI4WCZGRkJJMnD+fkyRNYWBgQFKRi7NjJDB48IoXN8uXl7Nhxis/BCF8z1t3f35+OHRvj4/OOokXVOjvt2nVm1aqUktMnThzj7t07WFhAWFgUx4/voUuXfplSYq1WrQb7919CpVJhYVFQ4+ZlTEw05uaFMTIy4rvv9AkLk2JqmjlVxvDwMNzcnqBQxFGunCkvX77G2/tVpg6VBQV9YPTofri63qJIEQOiow2YNWuZ1jlwM8ro0UM4enQPhQtDUBCULGnBkSPXcpW+UVbM6NcCToIglAecgHVZYDNTrF69jDt3tvD8eRSCEMbz51HcvbsFJyexMqGv71vGjh3IihVxfPigHqRFi2IZObIv/v7+ororVy7g8ePtCEI0ghDG06dRXL++gQ0bnLRu55Ils3Fz242bWzQvXoTy6FEkly6tYcuW9Rqv0dPTS/cPS3qdV0ZspuXkfX3fMmBAJ5Yu9cXbOwJf3yg6dnxKjx5tUCgUad5fE97er/njj244Ovrx+rXa5i+/PKJnz7YpwkCnTx9NZORJXr2Kxs0tjBs3wtm69R9OnDim0b5EIvnqB5o6dmxMrVrvCAiAt2/V+VWvXj3IkiXzRfVOnTrBhg0LOXlSxbt36qXFHj1C6dr15xR914Yv9XvkyD5YWDjz5k00Hh7hnDkTwty5Y7l500Xrezo49KRMmWu8fRuNp2c4x49/ZNaskZnSgbK370b58jfibYZx6FAwU6YM5cGDe1rbTC+Ojsu4cGEPN2+qxzIgABo0CKBDhx+y/d4ZIVPhlXK53AJwA8wFQVDI5XIp6lm9rSAIgWlfnX3hlbVqWXLyZBDVkuRRePwY2rYtyv37rxLKHByGEBGxhxPJ0pS2bAkWFn1YtmwVoF6KqFatNFeuhFAxiSrv3bvQtasFrq7JlXW+jFKppHLlkri6hmNtnVh+4wYMGFCSa9dyT/RHelm2bD6fPi3GyUkcKVKvXgHGjt3Kzz+30HClZhYsmEVcnCPLlyfaVKmgZk0zpk3bk5DcISwslBo1rPDyihZFihw5AkuXVufwYe1US5OSFaF7Hh7uNGtWC39/cQ7gnTth8uSC3Lv3JqGsRYvadOjgxsyZifWUSihZEmbO3ETHLEi4o6lPb9++oUWLmvj4RJFUWn3tWjh3rpVWy1yvX7+ibdu6+PhEkfRs1sqVEq5fb4eT064M2/TwcKdTpx/w9o4UCQkuXizh4cNOLFuWvYl56tYtw/TpH+nfP7EsPFydJ/rUqRuZzoEBWRNemdmlm9LAW0EQFADxzt43vvxLjh7gc4O1olixAqmW+/l9pHyyJT+5XF2e9JrAwHcpwhsBqlaFp0/fJdRVKpUEBISkavPdu2CN7UiL6OhoQkIiKZfs/I+6nR+0spnTfPrkh1wek6K8QgUl4eEfterTp09+1KoltimRQIUKEBGRaDMyMhgzM2mKcEC5XL1UklWfZ2btuLr6UqSI2MmDup1RUREi++Hh70UTC1CHwNrYgJ+fd7b26dWrMEqXNsTYWLzPUaEC7Nrlq9W93d1DsbQ0xNAwuU0Vx45pZ/PZsxCsrQ3R149MYfPSpbfZ/hxFRoanCBM2NYVixSAgwIemTRukfmEGyWw/Mr1Gn1myY0Zfp05ljh17QrduiWVHj0LdupVF1zRq9CN79lxhzhwSMtjHxsLBgzB48E+iurVqVeDYsRciqeKjR6FOnapaz/IqVrTk9Gkv2iU5l6W2WSNPHvqoXLkBBw/uZ8SIRFnhiAg4d07J4MHafU5VqjTiyJFjDBmSaDMsDC5dimP06CoJNg0NC6KnZ8ydOxHUSZL69MgRPWrVapAln2dWzOgrVvyejx/h2TOonORc1qFDUKJEWZF9K6va7Nt3XvQ9/vBBnYN49uxW2dqnYsVK8+qVOjmHVZLI3MOHDalRo7FW9y5RoixubjF4e6s3/z9z5IghNWo00cpmyZJWPH0aja8vfJfkoOuRI0ZUq6adzYzw7bfWHDr0goYNE8sEQb2EU6NGvWwZoyQz+nQjnZn0vTCDrFq1SgFMWbVq1aIRI0ao4pdu5gBzRowYEfGFywsDoyMjY1JkcE8PpqZGRESknD0CWFpWwN7+MEZGCgwM4MABCRMnmrBw4QbRBl6tWnVwcnLi/PkYSpVSyxAPGQIfPhRi3bpdojXM0qVtsbc/iomJAn192LtXwtSpJixevFlr5btSpayxtz+BqWkcenrqTDszZ8pYsmRTnjyebWNjy9ath3B2DqV4cQXPnsHgwTK+//4XevVKecQ+fTbLs2HDflxcwrGwUPD4MQwaJKNBgw5065Yoqqanp4e5eXGGD79E0aJxxMXB+vVSVq2SsWLFFooWzbykclrfufRiZGSEh4c7f//9nBIl1CGGjo6wahWsXr0nIekJQL16DZg+fS0eHiosLNRLhXZ2YGlZnVGjsiYMVVOfDA0NMTQ0ZPx4VywsYomIgGXL9Dl4sABLl27AzCzjM8zP+jwTJtylePFYwsJg8WJ9jh0rxJIl6zE1zfgZCmNjE2Jjo5ky5QHFi8cSEgILFhhw9mxhlixZl+1CgpUq1WDChB0EB6tzGly9Cj17wo8//krXrj2/bCAdJB8jiUSCTGYIsALQfDQ9CZmWQJDL5c7ARkEQdsrl8l7AQEEQfkrHpZZkowTCo0cPWLNmAe7uz7G1rcTQoRNSFU76+PEjDg4DefRIrVlbs2YzHB03ULBgSn3sBw/usWbNAjw9XyCXV2Ho0IlaJ1T+zN27t1m3biGvX7tTvnxVhg6dmCXrejlFWFgoa9eu5OLFQ5iamvLrr/3o1auf1kqToM7vunbtCi5dOoJMZkqnToPo0aNPqpu4//7rzMaNi/H19aZ69foMHz4BKyvrVKxmnKw8Xr9kyXx2715FVFQEJUqU5Z9/nKhfv2GKeh4e7owePZCXL59iYGBAy5Y9mD9/Saby5SblS306c+YU27cvJzDQn3r1fmb48D/57ruSmbrnyZPH2bFjBUFB72nQoDnDho3PVISKSqXixImj7NzpSEhIMPXrN2fo0PEUL67dwa6McueOK5MmDcPP7yXGxiZ07TqUCROmZpn9XJFKUC6XV0AdXlkECEYdXpmeYFJLcljrJjeR3/oD+a9P+a0/kP/6lN/6A7ljMxZBEF4AqWxp6tChQ4eO3EC+PhmrQ4cOHTp0jl6HDh068j06R69Dhw4d+ZycjKOXgnpjQVsyc21uJL/1B/Jfn/JbfyD/9Sm/9QfEfUry93SHsuVkhqkfgMyfS9ehQ4eO/08aA9e/WIucdfRGQB3AD9CseKVDhw4dOpIiBb4F7gDR6bkgJx29Dh06dOj4Cug2Y3Xo0KEjn6Nz9Dp06NCRz9E5eh06dOjI5+gcvQ4dOnTkc3SOXocOHTryOTpHr0OHDh35HJ2j16FDh458To6nEswocrm8PGr9e3PUicj7CILgnrOtyhxyufwVEBX/B2CCIAjncqxBGUQuly8GOqHOMVBVEISn8eV5dqzS6NMr8uBYyeVyc2AHYA3EAO7AEEEQAuVyeX1gHWCCWt+8lyAIATnV1vTwhf6ogCeAMr56b0EQnuRMSzOGXC4/ilpnXgmEASMEQXiY2a++mIQAAAMLSURBVGcpL87o1wJOgiCUB5xQf0HzA50FQagR/yfXO45kHAWaAK+TleflsdLUJ8ibY6UCFgqCIBcEoSrgCcyXy+V6wE5gePw4/QvMz8F2ppdU+5Pk/zdMMkZ5wsnH01cQhOqCIHwPLAY2x5dn6lnKU45eLpdbADWBPfFFe4Cacrm8WM61SocgCNcFQfBJWpbXxyq1PuVlBEEIEgTBOUnRLaAsUAuIEgThs2bKWqDrV25ehkmjP3kaQRA+JflnIUCZFc9SnnL0QGngrSAICoD4//rGl+d1dsnl8sdyuXy1XC4vnNONyQJ0Y5VLiZ/FDwWOA2VI8tYiCMJ7QE8ulxfNoeZlmGT9+YyzXC5/KJfL58nlcqMcappWyOXyjXK53Bv4B+hLFjxLec3R51caC4JQHbXImwRYlcPt0aGZ/DBWjqjXf/Ni21MjeX/KCIJQG/XSWyVgWk41TBsEQRgkCEIZYDKwKCts5jVH7wOUlMvlUoD4/34XX55n+bxEIAhCNLAaaJSzLcoSdGOVC4nfZLYFugmCoAS8SbLkIZfLvwGUgiAE5VATM0Qq/Uk6RiHARvLYGH1GEIQdwE/AGzL5LOUpRx8fCfAQsIsvsgMeCIIQmHOtyhxyudxULpcXiv+7BOiOuo95Gt1Y5T7kcvlc1GvyHeJ/qADuASZyufyH+H/bAwdyon0ZJbX+yOXyInK53CT+7/pAZ/LIGMnlcjO5XF46yb/bA0FApp+lPCdTLJfLK6AOMyoCBKMOMxJytlXaI5fLrYBDqDWmpcBzYKQgCH452rAMIJfLVwIdgRLAe+CDIAiV8/JYpdYnoD15dKzkcnll4CngBkTGF78UBOF3uVzeEHUUhzGJ4ZX+OdLQdKKpP8BC1H1RAQbADWC0IAhhOdHOjCCXy4sDxwBT1Dk6goDxgiDcz+yzlOccvQ4dOnToyBh5aulGhw4dOnRkHJ2j16FDh458js7R69ChQ0c+R+fodejQoSOfo3P0OnTo0JHP0Tl6HTp06Mjn6By9Dh06dORzdI5ehw4dOvI5/wMdYzqeOdGEzgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def form_linearly_separable_data(n=500, x1_min=0, x1_max=30, x2_min=0, x2_max=30):\n",
    "    data, target = [], []\n",
    "    for i in range(n):\n",
    "        x1 = np.random.randint(x1_min, x1_max)\n",
    "        x2 = np.random.randint(x2_min, x2_max)\n",
    "        if np.abs(x1 - x2) > 0.5:\n",
    "            data.append([x1, x2])\n",
    "            target.append(np.sign(x1 - x2))\n",
    "    return np.array(data), np.array(target)\n",
    "\n",
    "\n",
    "X, y = form_linearly_separable_data()\n",
    "plt.scatter(X[:, 0], X[:, 1], c=y, cmap=\"autumn\", edgecolors=\"black\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "训练一个决策树对上面的数据进行分类，并绘制分类边界。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Easy task. Decision tree compexifies everything')"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tree = DecisionTreeClassifier(random_state=17).fit(X, y)\n",
    "\n",
    "xx, yy = get_grid(X)\n",
    "predicted = tree.predict(np.c_[xx.ravel(), yy.ravel()]).reshape(xx.shape)\n",
    "plt.pcolormesh(xx, yy, predicted, cmap=\"autumn\")\n",
    "plt.scatter(\n",
    "    X[:, 0], X[:, 1], c=y, s=100, cmap=\"autumn\", edgecolors=\"black\", linewidth=1.5\n",
    ")\n",
    "plt.title(\"Easy task. Decision tree compexifies everything\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "可视化决策树。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1a584d9902214f6f883b3eb69a7d19d2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Image(value=b'\\x89PNG\\r\\n\\x1a\\n\\x00\\x00\\x00\\rIHDR\\x00\\x00\\x08K\\x00\\x00\\x04\\x1c\\x08\\x06\\x00\\x00\\x00gI\\x87\\xf3\\x…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "dot_data = StringIO()\n",
    "export_graphviz(tree, feature_names=[\"x1\", \"x2\"], out_file=dot_data, filled=True)\n",
    "graph = pydotplus.graph_from_dot_data(dot_data.getvalue())\n",
    "Image(value=graph.create_png())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，决策树构建的边界过于复杂，而且树的深度过深，产生了过拟合现象。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "再训练一个 k-NN 模型，看看它在这个任务上的表现情况。该单元格执行时间较长，建议在线下练习："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<pre style=\"font-size:14px; line-height:17px;\" class=\"hljs\">\n",
    "knn = KNeighborsClassifier(n_neighbors=1).fit(X, y)\n",
    "\n",
    "xx, yy = get_grid(X)\n",
    "predicted = knn.predict(np.c_[xx.ravel(), yy.ravel()]).reshape(xx.shape)\n",
    "plt.pcolormesh(xx, yy, predicted, cmap='autumn')\n",
    "plt.scatter(X[:, 0], X[:, 1], c=y, s=100,\n",
    "            cmap='autumn', edgecolors='black', linewidth=1.5)\n",
    "plt.title('Easy task, kNN. Not bad')\n",
    "</pre>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![图片描述](https://doc.shiyanlou.com/courses/uid214893-20190505-1557037780636)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，最近邻方法的表现比决策树好一点，但仍然比不上线性分类器 $x_1 = x_2$（线性分类器将是下一个实验的内容）。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  在 MNIST 手写数字识别任务中应用决策树和 k-NN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "现在可以看看这两个算法应用到实际任务上的表现如何，首先载入 sklearn 内置的 MNIST 手写数字数据集，该数据库中手写数字的图片为 8x8 的矩阵，矩阵中的值表示每个像素的白色亮度。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.,  0.,  5., 13.,  9.,  1.,  0.,  0.],\n",
       "       [ 0.,  0., 13., 15., 10., 15.,  5.,  0.],\n",
       "       [ 0.,  3., 15.,  2.,  0., 11.,  8.,  0.],\n",
       "       [ 0.,  4., 12.,  0.,  0.,  8.,  8.,  0.],\n",
       "       [ 0.,  5.,  8.,  0.,  0.,  9.,  8.,  0.],\n",
       "       [ 0.,  4., 11.,  0.,  1., 12.,  7.,  0.],\n",
       "       [ 0.,  2., 14.,  5., 10., 12.,  0.,  0.],\n",
       "       [ 0.,  0.,  6., 13., 10.,  0.,  0.,  0.]])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.datasets import load_digits\n",
    "\n",
    "data = load_digits()\n",
    "X, y = data.data, data.target\n",
    "\n",
    "X[0, :].reshape([8, 8])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "绘制一些 MNIST 手写数字。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, axes = plt.subplots(1, 4, sharey=True, figsize=(16, 6))\n",
    "for i in range(4):\n",
    "    axes[i].imshow(X[i, :].reshape([8, 8]), cmap=\"Greys\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用 `train_test_split()` 方法分割数据集，其中的 70% 作为训练集（X_train，y_train），30% 作为留置集（X_holdout，y_holdout）。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_holdout, y_train, y_holdout = train_test_split(\n",
    "    X, y, test_size=0.3, random_state=17\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用随机参数训练决策树和 k-NN。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Pipeline(memory=None,\n",
       "     steps=[('scaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('knn', KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',\n",
       "           metric_params=None, n_jobs=None, n_neighbors=10, p=2,\n",
       "           weights='uniform'))])"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree = DecisionTreeClassifier(max_depth=5, random_state=17)\n",
    "knn_pipe = Pipeline(\n",
    "    [(\"scaler\", StandardScaler()), (\"knn\", KNeighborsClassifier(n_neighbors=10))]\n",
    ")\n",
    "\n",
    "tree.fit(X_train, y_train)\n",
    "knn_pipe.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "训练好之后，分别在留置集上做出预测。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.975925925925926, 0.6666666666666666)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_pred = tree.predict(X_holdout)\n",
    "knn_pred = knn_pipe.predict(X_holdout)\n",
    "accuracy_score(y_holdout, knn_pred), accuracy_score(\n",
    "    y_holdout, tree_pred\n",
    ")  # (0.976, 0.666)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，k-NN 做得更好，不过别忘了我们用的是随机参数。现在，使用交叉验证调优决策树模型，因为这次任务所需考虑的特征比之前任务中的更多，所以可以增加参数的大小。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 9 candidates, totalling 45 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 2 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  45 out of  45 | elapsed:    0.4s finished\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/model_selection/_search.py:841: DeprecationWarning: The default of the `iid` parameter will change from True to False in version 0.22 and will be removed in 0.24. This will change numeric results when test-set sizes are unequal.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise-deprecating',\n",
       "       estimator=DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=5,\n",
       "            max_features=None, max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, presort=False, random_state=17,\n",
       "            splitter='best'),\n",
       "       fit_params=None, iid='warn', n_jobs=-1,\n",
       "       param_grid={'max_depth': [10, 20, 30], 'max_features': [30, 50, 64]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score='warn',\n",
       "       scoring=None, verbose=True)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_params = {\"max_depth\": [10, 20, 30], \"max_features\": [30, 50, 64]}\n",
    "\n",
    "tree_grid = GridSearchCV(tree, tree_params, cv=5, n_jobs=-1, verbose=True)\n",
    "\n",
    "tree_grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "查看交叉验证得到的最佳参数组合和相应的准确率。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({'max_depth': 20, 'max_features': 64}, 0.8440731901352426)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_grid.best_params_, tree_grid.best_score_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "调优后决策树模型的准确率达到了 84.4%，但还不到使用随机参数的 k-NN 的准确率（97%）。现在，使用交叉验证调优 k-NN 模型。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9865458422397149"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(cross_val_score(KNeighborsClassifier(n_neighbors=1), X_train, y_train, cv=5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，调优后的 k-NN 在这一数据集上可以达到 98.7% 的准确率。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面在这一数据集上训练随机森林模型，在大多数数据集上，它的效果比 k-NN 要好。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9349085619110404"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(\n",
    "    cross_val_score(RandomForestClassifier(random_state=17), X_train, y_train, cv=5)\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上可知，在这个数据集中随机森林的准确率（93.5%）不如 k-NN（98.7%）。当然，我们没有对随机森林的参数进行任何调优，但即使经过调优，训练精确度也无法超过 k-NN。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "决策树、k-NN、随机森林在这个数据集上的准确率如下所示："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "|   算法\\方式      |   留置法  | 交叉验证 |  \n",
    "|---------|-------|---------|\n",
    "| 决策树  | 0.667 |  0.844  |  \n",
    "| k-NN | 0.976 |  0.987  |\n",
    "| 随机森林  | / |  0.934  | "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从这个任务中得到的结论（同时也是一个通用的建议）：首先查看简单模型（决策树、最近邻）在你的数据上的表现，因为可能仅使用简单模型就已经表现得足够好了。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 最近邻方法的复杂情形"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面考虑另一种情况，即在一个分类问题中，某个特征直接和目标变量成比例的情况。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "def form_noisy_data(n_obj=1000, n_feat=100, random_seed=17):\n",
    "    np.seed = random_seed\n",
    "    y = np.random.choice([-1, 1], size=n_obj)\n",
    "    # 第一个特征与目标成比例\n",
    "    x1 = 0.3 * y\n",
    "    # 其他特征为噪声\n",
    "    x_other = np.random.random(size=[n_obj, n_feat - 1])\n",
    "    return np.hstack([x1.reshape([n_obj, 1]), x_other]), y\n",
    "\n",
    "\n",
    "X, y = form_noisy_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用最近邻方法训练模型后，查看交叉验证和留置集的准确率，并绘制这两个准确率随 n_neighbors 最近邻数目 参数变化的曲线，这样的曲线被称为验证曲线。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f5371c73048>"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAELCAYAAAA2mZrgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3Xd8VNeZ+P/PvVM0qqiijkQ9NNMxBgMGA+4tjhtxS92vU5zNpse768TJ2nESb7K/JE7s2HFixw5xS1xxBdvE2MSmGQPWoUoCgYSQRhJq039/zEgIgdBISBrNzPN+vfSS5pa5z5kZ3WfOPeeeYwQCAYQQQggAM9IBCCGEGD4kKQghhOgkSUEIIUQnSQpCCCE6SVIQQgjRSZKCEEKITpIUhOiFUqpcKbW8H/stUUodHIyYTnPMRKXUi0qpRqXU02Fsv0MptST094+UUo8PepBiWLNGOgARm5RS5UAu4Ouy+M9a668N0fF/BIzTWt80FMfrq9CJ+C3g91rrr3RZ/i7wsNb6z0qpzwJ/Ar6ntf55l20OAjdprd8+xVNfQ/B1z9Jae3uLQ2s95UzKIWKP1BTEYLpca53S5WdIEkIUaQFuVkqVnmabeuC7SqnUMJ+zBNgVTkIQ4lSkpiCGnFJqLPAQMB0IAK8BX9VaN4TWfw/4OpAGHAK+AuwA9gHFWuu60HazQvsWaK09XZ7/IuAOwFBKXQXs1VpPV0p9DvguUATUAj/TWj8Y2icb+DOwEPCHjnee1trfLfZJwGrgDq31qj6W++vAbcAFoUUNwD+AHwKf62G3TwAn8E3grl6e/y7gBxwv978Db3P617oc+KLW+s1uz+UAHgYuBizAbuAyrXVNuOUV0UlqCiISDOCnQAEwCSgGfgSglFLA14C5WutU4EKgXGtdTfAEd12X57kZ+FvXhACgtX4VuAd4MlRDmR5adQS4jGCy+Rzwq1BiAfgWcBDIIXj55Q6CJ9FOXZLQ7f1ICHcCnyWYaLq2M9wNfDpU7p78N/ANpVTm6Y6htf4hJ5b7j5zmte7FrcCI0PZZBJNZWxj7iSgnNQUxmJ5TSnW9jPEdrfVDWus9wJ7Qslql1C8JfluGYBtEAjBZKVWrtS7vsv+jBGsQv1dKWYCVwBXhBqO1frnLw3eUUq8Di4DNgAfIB0pC8f2z2+6LgC/Q87X8nhih8p0NLNVaN3aLqVop9QDwY+D6HuLeqpR6A/he6CdsvbzWp+MhmAzGaa23AZv6clwRvSQpiMF0VffLEgBKqVzg/yN4ok0lWGN1QvAkppT6BsFvs1OUUq8B39RaHwKeBx5QSo0GFNCotf4g3GCUUhcTPCFOCB0zCfg4tPoXoWO+HvrS/get9b1ddr8NeKePCQEgHfg34PruCaGLnwF7lVLTe1gPcCfwQeikHrbTvda9+AvBWsLflFLpwOPAf3avlYnYI5ePRCTcQ/DSzFla6zTgJoKXOQDQWv9Va72QYKNpgOBJE611O/BUaPubCZ64etL90k8C8CxwH5CrtU4n2DZghJ77mNb6W1rrMQRrH99USi3r8hS3AaOUUr/qY1mdBC9Z/Ukpde6pNgi1kfwf8JOenkRrXQb8HfjPPh7/tK/1aY7n0VrfpbWeDCwIleGWPh5bRCGpKYhISAUagUalVCHwnY4VoWvrhcB6oJ3gdWxLl30fC/2MJHjdvyc1wAqllBlqLLYTvCxVC3hDtYYLgO2h414GlAF7Q7H5CDY4dzgGXASsUUrdq7X+friF1Vq/rZS6Efi7UuryHmo3vyTYkH66E/ZdwLZetumux9f6dJRSS4GjwE6gieDlJP9pdxIxQWoKYjC9qJRq7vLzj9Dyu4BZBE9WLxP8BtwhAbiX4AmpmuDJ/wcdK7XW6wmenDZrrStOc+yOG7fqlFKbtdbHCLZHPEXw2/tngBe6bD8eeBNoBt4Hfqe1fqvrE4Z67KwALlZK/QRAKfWKUup0yalj3zeAz4dek1mnWN8E/BzosTFZa72fYO0oubfjdXG61/p08oBnCCaET4B3OH3NTMQIQybZEdFGKbUW+KvW+uFIxyJErJHLRyKqKKXmEvzme2WkYxEiFsnlIxE1lFKPErzE843Q5SAhxACTy0dCCCE6SU1BCCFEp2hoU0gA5gKHOXHETSGEED2zELxL/0PAFe5O0ZAU5nLykANCCCHCswh4N9yNoyEpHAZwOlvw+/ve/pGVlUJdXfOABzWcSZnjg5Q5PvS3zKZpkJGRDKFzaLiiISn4APz+QL+SQse+8UbKHB+kzPHhDMvcp8vu0tAshBCikyQFIYQQnSQpCCGE6CRJQQghRCdJCkIIITpJUogxMmyJEOJMREOXVBEmb8UW2tf9if0+DySkYDhCPwkpGI7U0OPU48u7/m3KR0EIIUkhJgT8ftwb/45760uYWaNIGTuN1vo6Aq5mAu3N+BsOE2hvBk97z09iTwwmiYTuCaOHZJKQjGFaen4+IURUkqQQ5fxtTbSvfQBf1U5sExeTsOAmsvOzqK09eWTpgM9DoL059HPs+G9Xt2VtjfidVQTaj4HX3fPBE5K71ERSMJPSMdPzMdPzMNMLMFKyMUy5QjmQAh4X3oMf4ztURrOaTiBzirzGYkBJUohivpo9tL35OwLtTTgWfx7bxMWn3d6w2DCSMyA5I+xjBLzuk5NIe3MokRw7nmRanHhr9hJwdbkd37RijsgNJYp8zBF5nX8b9sT+Fjvu+Fsb8VZuxVu+GV/VDvB5wbBwZMebGGm52KddhG3CuRhWe6RDFTEgrKSglJoAPApkAXXALVrr3d22yQMeBEYDNuBurfXjoXUW4NcEJz4PAPfKVIr9FwgE8Oxcg+v9VRjJmSRd+d9YsksG5ViG1Y6RkgkpPU4dfGJsoctV/obD+BoOE2isxld/EG/5Zggcn/fd6KxVdEsWKZkYhnzz9TdU4ynfjLdiM/6avUAAIyUL26SlWEtnYckdR1KD5ui6Z3G9+yjuTf/ANvUC7JOXYiT0ZQpnIU4Ubk3hAeB+rfXjSqmbCJ78z++2zS+BjVrrK5VSOcAmpdQ7WusDwI3AOIKTo2cBW5RSb2qtywekFHEk4HHR/s8/493zPpZR00lc+m/D6iRgOFKw5I3HkjceW5flAZ8X/7EjnQkj+FONZ88GcLce39Bix0zPxRyRfzxphBKHYUsY8vIMlUDAj//IPrwVW/CWb8bfEBzDzMwqwT77SqylszAzizEMo3OflInzac2cgu9wGe6tL+P+8BncW1/CNmkJ9rMuxOxDjVCIDr0mBaXUSIJz4q4ILVoF/FYplaO1ru2y6XTgVwBa61ql1FbgOuB/geuBh7TWfqBWKfUccC3wiwErSRzwN1TT9sZv8DsPYZ9zNfaZl0XNt2rDYsWSXoAlveCE5YFAgEBbE/7G6hMShq92P959HxKsWIaeIzmzS5tFPmZ6QTBZROnJL+B14zv0Cd7yLXgrthBoawTDgqVAkTD5fKwlMzFTs0/7HIZhYC2YhLVgEr6jFbg/egXPx6/h2f4GtvELsE2/+KTXXIjTCaemUAxUaa19AFprn1LqUGh516SwCbhBKbURKAUWAOWhdaOAii7bVob2F2Hy7N9I+9sPY5hWEi/5FtaiqZEOaUAYhoGRNAIzaQTkqxPWBbxu/E3daheN1Xh2rT+xJ5XNgTurEH9Kzom1i7TcYXedPeBqwVv5Ed7yzXgPbg+Ww+bAWnwW1pKZWEdN73fNz5JdQuKy2/DP/TTuba/i0evw6Hexls7EPv0SLLnjBrg0IhYNZEPztwjWFLYSPOmvAbwD9eRZWSn93jcnJ3WgwhhyAb+P+rcep33DCyQUjCf36m9hHZHT637RXOYT5GcBk05YFAgE8DU78dRV4T5ahaeuCk99Ff4je3Dv2dBlSwNreg62rEJsWYXYQ79tWQVYktNPuBQzmDyNR2jd9SEtuz6kvWIHBPxYktNJnbqY5AlzSSw9C8Nq6/2JTuGU73NOKoz9Cr6WG2n8cDVNm16ltXwzjlFTSJ9/FYljZw5Z2QdDzHy2+2AoyxxOUjgAFCqlLKFaggUoCC3vFLqUdFPHY6XUamBn6GElUEJwWjg4uebQq7q65n6NKZ6Tk3rK7pnRwN/aQPua3+M7rLFNPh/b/JU43TbopTzRXObw2SC5NPhTAvmhMgc8rhMvRTVW42o4TFv5DvB16V5rTwy1VXRru0gbiWE5s+9KgUAAf13l8faBukoAzPQC7NMvwVo6EzNnNBgmLUCLsx04zT0kPej9fTZhymUkTViO55N3cH38GtVP3o2ZWYx9+sVYx86LuntN4uOzfaL+ltk0jX59me7106+1PhJqH1gJPB76vaVbewJKqSygUWvtVUqdD5wFXBNa/TTwJaXU3wk2NF9FcIo40QPvYU37m78j4G7DseRL2CacG+mQooJhS8CSXXJSb6xAwE+guf7ktouqHXh3r+/yBCZGWk4oWRxvu7Ck52M4ev4HC/i9+A7v6kwEgeY6wMCSO46EeddhLZmFmZ43SKU+PcPmwD7tQmxTluHduwH3R6tpf+sPGB8+G+zOOnExhjV2G/FF34T7leg24FGl1J2AE7gFOmsDd2qtNwJnA79WSvmAo8DlWuuObiV/AeYBHd1Yf6y13j9AZYgpgUAAz8ev4/rXkxipOSRd+m0smdL8cqYMw8RIzQ423HZrjwm42zprFSf0jDq4HfzHr4AaCSldahXBhBHweYPtAwe2gasFLDYshVOwzroC66gZwbaSYcKwWLFNWIh1/AJ8lR/h3roa13tP4N70PLapy7FPWX7axCfigxEFA6iVAvvj4fJRwN1G+7pH8O77EGvpLBxLvohhT+rz80RTmQfKYJQ54PcTaD7arRttMHkE2pqOb5iQjLVkBtaSWViLpg5Z19mBKLO3ejfurS/jq9wKVju2iedhn3YRZkrWAEU5sOSzHb4ul49Gc7zTT6/kjuZhwuesov313+BvqsF+9nXYp18c1Y2Bg8UfCHC0oY2qoy0cCv1U17disVowAgHsVhOb1YLNap78YzGx2yzYLCcut/ewvd1qwWbLwJafha1oGqZ5/P0ItDcHk0MggGXkmBOuzQcCATxePx6fH7cn+Nvj8XV73LHe18vjbvt7/Xi8fgKBADdcMJGJhWln9Hpa88Zjvegb+OqrcG9bjWfHWjw71mIdNy/YYymz6IyeX0QfqSkMA549G2hf9ycMWwKOZV/GWjCp9526aHd72XWggZ3lTnYfbCA1OYH0ZDs56Q5GZiQFf6cnkuToXw+XSOg4+R862krV0eZQAmjlcF0Lbu/xO6MzUhPIz0oiKdFOc4sreAL1nvjTcSL1+vynOWLvLKaB3WZ2SSoWDIOTjnWmx7FaOpJSz8mrrqmdQ0dbuOkCxdKZhWd0vK78zXW4P34dzydvg9eFZdT0YHLImzAsvqREw//zQJOaQhwJ+Ly4/vUknu1vYMkdj2P5V8K6C9Xn91N++Bg7yuvZWe5kb1UjPn8Aq8VkbEEaLo+PrbtraWr1nLBfssNKTnoiOemJjMxI7Pw7J91BZqrjhG/CQ8UfCHC0sZ1DtS2hk38rh462nPLkX5CdzJJRhRRkJ1OYnUx+VjJJjuBHOJx/HH8ggLf7N3ivH4/Xd2Ly8Ppxh5adKrl0/ebu9wc6ayanO4mf+vHx5NLx2Go1McM4+bo9Pv74Shl/eU3T2u7hknNKBuSkbaZk4Zi/koSZl+PeuQbP9jdpe/GnmLnjgj2nSmZEzQ2Ton+kphAh/hYnbW/ej79mD7apF5BwznU9zmkQCASorm9lZ7mTneX1lFU20ObyYgCjclOZPDqDyaWZjC8cgd1m6Sxzm8vL0cZ2jjjbqG04/nOkoY26xnZ8XV5Pi2mQPcJBTihZjEw//js73YHDfmbfH/p68i/MTqag46fLyb8nw/V9HkwZmcn87M8fsGFnDRfNG8W1S8YO+Lf5gNeFR/8T97ZXCRw7iplegG3ahdjGzsOwOQb0WOGIx/dZagpxwHvoE9rX/J6Ax4Vj2ZexjZ130jZNLW52VtSzc7+TnRX11De5AMge4WDuxJFMLs1gUkkGqUk937GbmGCleGQKxSNP7lHi8/txNrk6k0RtQ3vwt7ONvVVNtLlOvO8wLXQ5qmvC6PhJT7F3now6T/6h6/1VtS2nPfmfV1xIYU7HyT8pqi5xRZrVYvLFyyeT6LDy6r8qaWnzcOtFEwe0xmdYE7BPWY5t0lK8+z7E/dHLuNb9Cdf7q7CNPRubWow5cuCTkYgcSQpDKBAI4P5oNe4Pn8EckUfiZd/HkhEcl8bl8YXaBYKXhA4cCQ5BneywMrEkg8vmZzK5NIOc9MQB+Qe0mCbZ6YlkpydyqhaM5jbPibWLUG1j94EG/rWjhq51Nrs1+Fw2i8nh+hbcnm4n/6wkzpsROvlnJVOQLSf/gWIaBjetmECyw8ZL75XT5vLypcunYLMO7CUew7RgG3cO1rHz8NfswV22Ds+eDXjK1mFmFGBT52GdsADTEX93G8caSQpDJOBupf2th/BWbME65mzsCz9LRb2XnZ+Us7O8nj1VjXh9AawWg3GFI/j0eWOYXJpJSW5qRK71pyTaSEm0MTr/5N4tHq+fuqb2E5JFbUMbbq+f84oLKchOojA7RU7+Q8QwDK5ePIZkh5Un1+6hzb2Nr33qLBLsA3+3smEYWPLGk5g3nsCCz+DZ9wGesndwbViF64OnsJbMxDZxMZbCqTL5T5SSpDAEfHUHgqObHjvKgeJLeLtxIp/8fiOtoUs0o0amsHx2MZNLMxhfnE6CbXgPPWCzmuRlJpGX2fd7KMTgufDsUSQlWPnzq2Xc9+QWvnHtdJIHMSkb9kTsE8/DPvE8fPVVePQ6vLvfw7t/I0ZyJja1CJtaiJna+1hdYviQpDCIjrW6qfpgDbm7n6XNb+eRYyvYV5dNZtoxZqkcppRmMqkkg7Tk4TWSp4hei6YXkOSw8uALO/jZE5v55vUzSE8Z/JvpLJmFWOavJHD2tXgrtuApewf35hdwb34BS+FkbGoR1tJZw27UWnEySQqDYE9VI6te28mc5rdY6NjFXl8eG3OuYsHcIr5YmkluxsC0CwhxKrPVSP79Wiu/ffZj7n18M9+6YQY56UMz/alhsWIbMxfbmLn4m+vw6Hfx6HW0r30AEpKDczyoxViyZOiW4Uq6pA6CPzy5nkXOf1BkHqVl7DKyz7sBaz+HRu4P6bYXH3or895DjfzfUx9htZp8+/oZFOZEZlyjQMCPr2onnrJ1wWlZ/V7MnNHBy0vjzunTUC7yPoevv11SpSVogPn8fsYefZs8SwOOFbeTt+zmIU0IQnQYWzCC7904C4B7n9jM3kONEYnDMEysRVNJXP4VUm76PxIW3Ag+D653H6P5L9+g7a2H8B7WRMEX1LggSWGAlVcfo9Sspi1jArbRsyMdjohzRTkp/OCm2SQ5rNy3ais7y+sjGo/hSME+dQVJn/4JSVfdiW3CArzlm2h78ae0PPV9XFtfxt/aENEY4520KQywvXsPMN9yDP+oiZEORQgARqYn8oObZvO/T27l/57+iP93xVRmq8j2CDIMA8vIMVhGjiHhnJV493+Ip2wd7g+exv3hs1hHTQ+2PYyaNuATAQV8XgKuZgLtxwi0n+r3ycvM1BwSL/w6ZtrIAY1lOJKkMMCaK8oASBnVt0HthBhM6SkJfO8zs/i/pz/id899zOcunsTCafmRDgsIToxkm7AQ24SFwXks9Do8u97FW7EFIykd24RzsalFmCNOnqQo4PcSaG/pcgLvcjJ3tZzyxH/C/N7d2RIxHCkYjlSMxDTMjEKMhGQ8u9fT+sI9JF7y7ZgfOVaSwgDy+vzYG/bjs1swc0ojHY4QJ0hJtPHtG2bw279/zCOrP6HN5WXF3OHVC8hMzyNh3nXY516Nt3JbsPbw0WrcW1/Gkjue6tQ02pucx0/w7raen8zmOH6Cd6Rgjsjr/Lvr8s7fCSk9TsVqm3gebat/QeuLPyXp4m9iGTl2kF6ByJOkMIDKDx+jxKzBlVaMYZHGZTH8OOxW/v2a6fzhhR2sWrOblnYPVy4cPey6SBumFVvpLGyls/C3OPHsWo9334d4m50Y9qTgXNqO1OCJ/FQneEfKgP4PWjILSbriP2ld/QtaX/o5iRd8HWvRlAF7/uFEksIA0uU1LLDUYy2eE+lQhOiRzWpy21VTePQVzQvry2lp97Jy+fiwhuyOBDM5g4SZl5Ew87KIdkk103JIuuIO2lbfR9urv8Kx7DZso2Pvf116Hw2ghnKN1fCTWKQiHYoQp2UxTT53yUQumFvMmk0H+eNLO894cqB4YCalk3T5DzBzSml/8348ZesiHdKAk6QwQDxeP7b6/QBYcsdHOBohemcYBtefP45PLR7D+ztq+N0/tuP2+CId1rBnJCSTdMl3sBROoX3dI7i3vRrpkAaUJIUBsv9wE6PMGlxJIzEckblzVIi+MgyDyxeUctMFE/hoz1F+9dRHJ82lIU5m2BJIvPDfsY6Zi2vD33B9+GzM3HwnSWGAlJXXM9paS0KBXDoS0ef8WUV86fLJ7Klq5OerttDU6o50SMOeYbHhOP/L2Caeh3vLi7jW/4VAIPovwUlSGCBHyveSZLpxSHuCiFLnTMnja1efxaGjLfzsic3UN52mP78AwDBNEhZ9Fvv0S/DsXEv7W38g4I/umpYkhQHg8fqw1O8FwJI3IcLRCNF/08dl883rptPQ7OKnj2+iur410iENe4ZhBO+tOPtavHs20Pb6bwh4XZEOq98kKQyAvVVNlJhH8NpTMWRCERHl1KgMvrtyFm6vn3sf30RlTXyNStpfCTMuJWHRZ/FVbqNt9f8ScEdnQpWkMADKKp2MsR7Bmjd+2N0EJER/lOSl8v0bZ2G1mvzsr5vZdUAGqQuHfdISHMtuw3dkL60v3ou/rSnSIfWZJIUBULn/AFmWZmlkFjElPyuZH9w4m7TkBH755Fa27T0a6ZCigm3sPBIv/Hf8DdW0vnAP/ua6SIfUJ2Hd0ayUmgA8CmQBdcAtWuvd3bYZCfwJKAZswFvA17XWXqXUj4CvAIdCm6/XWn91QEoQYS6PD7NuHySDJU/uTxCxJWuEgx/cOItfPrWV3zz7MV+8bDLzJudGOqxhz1o8jcRLv0Pbq7+k9fm7Sbz021jSCyIdVljCrSk8ANyvtZ4A3A88eIpt7gA+0VpPA6YBs4Gru6x/TGs9I/QTEwkBYG9VI6WWGvymDTN7VKTDEWLApSXb+e7KWYwtSOMPL+zg9Q8q5e7nMFjzxpN0+Q/A76XthZ/iqy2PdEhh6TUphGoAs4BVoUWrgFlKnTQgewBIVUqZQAJgB6oGMNZhKdieUItl5BgMU4aSErEpyWHlP66fwbSxWfxt7R5+8OD7vLHxAC633AF9OpasUSRdcQdY7bS+dC/eQ2WRDqlXvc7RrJSaTfBb/pQuy3YCN2mtN3dZlgk8C0wGkoHfaq2/H1r3I+CLQD1QDfxQa/1+mDGWAvvD3HbI3fHrNXz+2O/JPPfTZC5ZGelwhBhUfn+AjWU1PLt2Nzv315OaZOeyhaO59NzRjEhJiHR4w5a3qY7Dq36M11nDyKu/RfKEuUN5+D7N0TyQX22vBbYBy4BU4BWl1DVa62cIXn66W2vtUUqtAJ5XSk3SWofdAlNX14zf3/fbyAdzVEWX24fn8B7MlACutFHDZkJxmdw8PkSqzKNzkvn29TPYc7CR1RsqWPW65tm1u1k0vYALzy4me0TioB07et9nO/ZLvofvlV9S88zPcSz5IrbxC8Las79lNk2DrKy+D7kTTpvCAaBQKWUBCP0uCC3v6nbgCa21X2vdCDwPLAXQWldrrT2hv98I7Tu1z9EOM7urGig1awhgYMkdF+lwhBhS44pG8PVrpvGTL85j7qSRvL2liu8/sIE/vLiDA0eaIx3esGM6Ukm69LtY8hXtb/0B9/Y3Ix3SKfWaFLTWR4CtQMe1kZXAFq11bbdN9wMXASil7MByYHvocWHHRkqpGQQvCekzjD3iyioaGGM7gpFRiGFPinQ4QkREYXYyX7h0Mj+7bT7L5xSxZfdRfvjIB/zqqY/Qlc6YGShuIBj2RBIv+g+sJTNxvfc4rs3PD7vXJ9zLR7cBjyql7gScwC0ASqnVwJ1a643AN4AHlFIfAxaCXVIfCu1/T6htwge4gZu11tUDV4zI0BV1LLUdxZa/KNKhCBFxmWkOblg2nsvPLWXt5ire3HiAn/11C2MK0rh4XgkzJ2QP24l8hpJhteNY8bXgsNsb/0GgvYWE+TdgGMPjtrGwkoLWugyYd4rll3T5ey+woof9b+1vgMNVm8uLu7YSe5pH7k8Qootkh43LF5Ry4dxi1n98mFc/qOT+f3xMXmYSF80bxfwpedisw+MEGCmGacFx3hdw2ZPwbH+dgLsFx+LPY5iWSIcm03H21+6DjZRajgAyqY4Qp2K3WVg6q4jFMwrYpGtZvaGCP79SxnP/3MeKucUsmVFIYkL8noIMwyRh/mcwHCm4N/6DdlcrjmVfxrDaIxpX/L4jZ6is0slY2xFIysBIyYp0OEIMWxbT5OxJucydOJKd5U5Wb6jg6bf28tJ7FSydWciKOUVx253VMAwSZl2JYU/C9d4TtL36KxIv+DqGffB6cPVGkkI/lZXXsyihFmveFBkET4gwGIbBlNGZTBmdyf7DTbzyr0pe2VDB6x8e4Nyz8rho3ihyM+Kzw4Z96gqMhGTa336Y1pd/TtLF34rYDI6SFPqhtd1L09FqUka0yPwJQvTD6Pw0vnLVVGrqW3n1g0rWf1zNuq2HmK1yuPicEkbnp0U6xCFnG78Aw55I25v30/riPSRe8h3M5IwhjyO+W3v6adfBBkZ3tCdII7MQ/ZabmcStF03kF1+ez8XnlLCj3MlPHt3IL1ZtYfv+umHXXXOwWUtmknjxt/A319P6/P/gb6wZ8hgkKfRDWYWTsbajYHNgZhZFOhwhot6IlASuWTKW+76ygGuXjuVQXQu/fPIj7vrzh3zwSQ0+f/wMwGctmETSZd8Hr5vWF+7G2zS0Q5ZLUuiHskonKvEolpFjh0UXMiFiRWKClYvnlfDz2xbw2Ysn4vb4eeD5Hdzxhw2s3XwQX5yMzmrJKSVsSlRaAAAgAElEQVTxih9gZhTiaxvau8OlTaGPmts81B6pJyvjKJY8uWlNiMFgs5osnl7Awmn5bNl1lFf+VcHjr++izePn0nnxMUS9Jb2ApMu+R0JOKgzheE9SU+ij3QcaKLHWYiDtCUIMNtMwmK1y+M+bZ7Ngah7Pr9tHjTM65z6OFpIU+uiTSifj7LVgmFhGjol0OELEBcMwuGbJWGxWgyfX7Il0ODFNkkIflVU0MDmpHjNrFIbNEelwhIgb6SkJXLdcsXXPUbbvj655j6OJJIU+ONbq5lBtE3n+arl0JEQEXLl4DCPTE/nbmj0yJeggkaTQB7sONFBkqccS8EpSECICbFYL1y8bx6GjLby1JeZn+40ISQp9UFbRwPiE4DQSMgieEJExY1w2U0ozeP6f+znW6o50ODFHkkIflFU6mZrixEjNicjt50KIYKPzDcvG0+728dw/h+307VFLkkKYmlrcVB1tpojDMvWmEBFWmJPC0lmFvL21Sqb+HGCSFMKkDzSQZTZj97ZIe4IQw8BVi0aT7LCx6s1dcTdG0mCSpBCmsgonyhFqT5CRUYWIuGSHjU8tGk1ZZQObTpoyXvSXJIUwlVU6mZ7WAPYkzIyCSIcjhAAWzyigKCeZp97ag9vji3Q4MUGSQhgaml0crmtllFGDJXfcsJlgW4h4ZzFNVi6fwNHGdl77oDLS4cQEObuFoazSSZLhIslVK+0JQgwzk0oymK1yeHlDBfVN7ZEOJ+pJUgiDrmxAJQZvq5f2BCGGn+uWjsPvh2fe2RvpUKKeJIUwlFU4mTWiAUwLlpzRkQ5HCNFNTnoiF80rZsOOGvYcbIx0OFFNkkIvnMdc1DjbKLUewcwuxbDaIx2SEOIULjmnhPQUO399cxd+6aLab5IUelFW4cSCj9S2Q9KeIMQw5rBbuXbpOMqrj7H+48ORDidqSVLoRVmlkwlJDRh+r4x3JMQwd87kXMYWpvHsO/toc3kjHU5UCms6TqXUBOBRIAuoA27RWu/uts1I4E9AMWAD3gK+rrX2KqUswK+Bi4AAcK/W+uEBK8UgKqt0clFmE7TKTGtCDHeGYfCZ5RP4yaMbeem9cq5dKkPS9FW4NYUHgPu11hOA+4EHT7HNHcAnWutpwDRgNnB1aN2NwDhgPDAf+JFSqvQM4h4SdY3t1Da0M8ZWizEiFzMxLdIhCSF6MTo/jXPPyuP1Dw9QUy9Td/ZVr0khVAOYBawKLVoFzFJK5XTbNACkKqVMIAGwAx0Dnl8PPKS19muta4HngGsHIP5BVVbpBAKktx3AkitdUYWIFtecNxar1eTJtTJ1Z1+FU1MoBqq01j6A0O9DoeVd/QSYABwGqoHXtNbrQ+tGARVdtq08xf7DTlmlk9KkVkx3C5Y8qYYKES1GpCRwxYLS4NSd+2Tqzr4Iq00hTNcC24BlQCrwilLqGq31MwPx5FlZKf3eNycntV/77T7YyIr8FnBCzuSZ2LP69zyR0N8yRzMpc3wIt8wrL57Eu9urefqdvSyaMwqrJXr71Qzl+xxOUjgAFCqlLFprX6jRuCC0vKvbgc9rrf1Ao1LqeWAp8AzBmkEJ8GFo2+41h17V1TXj9/e973FOTiq1tcf6vF9tQxtHnG2Mzq7BcKTS4EvF6MfzREJ/yxzNpMzxoa9lvva8sfz62W089VoZK+YO+4sTp9Tf99k0jX59me41dWqtjwBbgZWhRSuBLaG2ga72E+xdhFLKDiwHtofWPQ18SSllhtoiriKYLIatsgonAJntB0OD4BkRjkgI0VfTx2UxZXQmz727nyaZujMs4danbgNuV0rtIlgjuA1AKbVaKTUntM03gEVKqY8JJpFdwEOhdX8B9gG7gQ3Aj7XWw3oevbJKJ/lJXiwtMgieENGqY+pOl0zdGbaw2hS01mXAvFMsv6TL33uBFT3s7wO+3M8Yh1wgEKCssoGluS3QiNy0JkQUK8xO5vxZhazZfJAlMwoYlRt/7TB9Eb0tL4PoSEMbzmMulOMoWKyYOaWRDkkIcQau7Jy6c7dM3dkLSQqn0NGekOOpwpIzBsNii3BEQogzkeyw8anFY9AHZOrO3khSOIWyygayk00sDQekPUGIGHHe9AKKclJ4cq1M3Xk6khS6CbYnOJmf3w5+n7QnCBEjTNPgM8vHU9fUzqsydWePJCl0U13fSmOzm0lJoZnWcuVOZiFixcSSDOaoHFa/L1N39kSSQjdllQ0A5HkPYWYUYDj6fye1EGL4uW7pOPwBeOZtmbrzVCQpdFNW4SQz1Y6lfr8MgidEDMpOT+SieaPYsLOG3QcbIh3OsCNJoYtAIICudHJ2vhfcrdLILESMuvScEjJSE/jrm7tl6s5uJCl0caiulaZWD1NTgl1SJSkIEZsS7BauXTKWiupjrN8mU3d2JUmhi477E/IDhzESR2Ckdp8yQggRK+ZNzmVc4QiefWevTN3ZhSSFLsoqnWSlJWCr24clb7wMgidEDDMMg5XLx9PU6uHF98ojHc6wIUkhxB8IoCsbmFloJdB8VC4dCREHRuensfCsfN748ADVMnUnIEmh06HaFprbPExLC/ZGkJvWhIgPnz5vDDaryZNrdkc6lGFBkkLIJ5XB9oRCqsFqx8weFeGIhBBDYURKApefW8pHe+v4WKbulKTQoazCSfYIB3bnfiwjx2KYAzlTqRBiOFs+u5iRGYn8bc1uvD5/pMOJKEkKBNsTdh1oYEpxMv66ShnaQog4Y7Oa3HD+eA7XtbJ2c1Wkw4koSQrAwSPNtLR7mZl+DAJ+LHlyJ7MQ8Wb6uCymjs7k+TifulOSAsfvTxhlVgMGltyxkQ1ICDHkTpi6c92+SIcTMZIUCA6CNzIjEbtzP2ZWEYY9KdIhCSEioCA7mfNnF/LO1kNU1hyLdDgREfdJwe8PoA80MKl4BL4je6UrqhBx7sqFo0lOtPHXOJ26M+6TQuWRY7S5vEzPbgNPu7QnCBHnkh02rl48hl0HGtgYh1N3xn1SKKsI3qw22nIEkEHwhBCwODR151Nrd8fd1J2SFCqd5GclYXfux0jOxEzJinRIQogIM02DG1eMp67Jxav/iq+pO+M6Kfj8fnYdaEAVp+Or2S21BCFEJzUqgzkTR7J6Q3xN3RnXSaGiupl2t4+puRBocUojsxDiBNctGUsAeDqOpu6M66RQFhrvaJwt2JgkNQUhRFfZ6YmsmFPMBztrOOKMj1FU4zspVDgpzE7G3rAfbA7MzOJIhySEGGaWzS7CNI24Gf4irFHflFITgEeBLKAOuEVrvbvbNo8B07osmgZcpbV+QSn1I+ArwKHQuvVa66+eYexnxOvzs/tgI+eelYevejeW3HEYZlznSCHEKWSkJjBn4kj+ue0QVy0ajcMe24Nlhlu6B4D7tdaPK6VuAh4Ezu+6gdb6lo6/lVLTgbXAa102eUxr/e0zjHfAlFcfw+XxMbnAgb/8IPbRcyIdkhBimFo+u4h/7azhve3VnD+rKNLhDKpevxorpUYCs4BVoUWrgFlKqdNNYPwF4AmttevMQxwcHeMdjXfUAQFpTxBC9GhMQRqj81NZs+kg/hi/yzmcmkIxUKW19gForX1KqUOh5Sfd7qeUsgOfAZZ3W3WDUuoCoBr4odb6/b4EmpWV0pfNT5CTk3rSsn2Hj1Gan0a66xANhknu5OmYdke/jzHcnKrMsU7KHB8iVeZPLR3PL/+6mSpnO7PUyCE99lCWeTAujl0FVGqtt3ZZ9gBwt9bao5RaATyvlJqktQ57mqO6umb8/r5n6JycVGprTxzYyuP1s3N/HYtnFHBs35uY2SXUNXoAT5+ffzg6VZljnZQ5PkSyzKogjbRkO8+u2UVxZuKQHbe/ZTZNo19fpsNpWT0AFCqlLACh3wWh5afyeeCRrgu01tVaa0/o7zdC+07tc7QDZP/hJtxePxOL0vAd2SeT6gghemWzmiyZUcC2vXXU1Mdu99Rek4LW+giwFVgZWrQS2KL1ySNFKaWKgEXAE92WF3b5ewZQCuh+R32GyiqdGMCElCbwuaU9QQgRliUzC7GYBms2HYx0KIMm3D6YtwG3K6V2AbeHHqOUWq2U6tpt51bgRa21s9v+9yiltiulPgIeAm7WWlefYez9VlbhpDg3BXv9fgC5k1kIEZb0lATmThrJux8fps3ljXQ4gyKsNgWtdRkw7xTLL+n2+O4e9r+1X9ENAo/Xx56qJs6fVYiv+l8YqTmYyRmRDksIESVWzClmw44a1n98mOVzYu+G17i7W2tvVRNen5+JMgieEKIfRuenMbYgLWa7p8ZdUiirdGIYMD7DQ6CtSS4dCSH6bNmcImqcbWzfVx/pUAZcHCaFBkpyU7E7gxNzy0xrQoi+mqNGMiLFzpubeuqEGb3iKim4PT72HWpkYkkGvurdYE/CzMiPdFhCiChjtZgsnVnI9n31HK5riXQ4AyquksKeqka8vgATRwWTgiVvPIYRVy+BEGKAnDejEKvFYO2m2Bo9Na7OiGWVDZiGwbhsC/6GQ9KeIITotxHJds6elMu72w/T2h473VPjLCk4Kc1PDc6fgEyqI4Q4M8vnFOFy+1j/8eFIhzJg4iYpuNw+9h9q6rx0hGnBkjM60mEJIaJYaV4a4wpHxFT31LhJCrurGvD5A0wsScdXswczuxTDao90WEKIKLd8ThFHGtrYtjfs8T2HtbhJCrqyAYtpMDYvCV/tPrl0JIQYELMm5JCeYmfNxtjonho3SaGswsno/DTsjQfB55WkIIQYEFaLydJZRewod1J1NPq7p8ZFUvD5/ew/fIwJxel4q4NTS0vPIyHEQDlvRgFWi8naGBg9NS6SQpvLhz8QYESKHX/NbowReZiJaZEOSwgRI9KS7JwzOZf12w/T2h7dk3XFRVJoDw1x67CZwZvWpJYghBhgy2YX4fb4+ee26O6eGh9Jwe0DYITPScDVjFXaE4QQA6wkL5UJRaHuqf2YOni4iIuk0OYO1hTSWisBuWlNCDE4ls8p5mhjOx/tPRrpUPotLpJCR00hqakcw5GKMSIvwhEJIWLRzAnZZKQm8ObG6G1wjqukYHfux5I7DsMwIhyRECIWWUyT82cV8kmFk6ra5kiH0y9xkRTaXF5SjTYsLbUyf4IQYlAtnl6AzWqyJkq7p8ZFUmh3+xhtrQWkPUEIMbhSQ91T39teTXNb9HVPjY+k4PIyxnoELDbM7JJIhyOEiHHLZhfh9vr557ZDkQ6lz+IjKbh9jLEdwZIzGsNii3Q4QogYNyo3FVWcztpNB/H5/ZEOp0/iIym43BRa6rHkjot0KEKIOLF8ThF1TS627o6u0VPjIin4XS1YDT9GckakQxFCxIkZ47PJSnOwZlN0jZ4aH0mhPThyoZGQHOFIhBDxwmKanD+7kLLKBg4ciZ7uqXGRFHC3AWDYkyIciBAiniyaVoDdakZVbcEazkZKqQnAo0AWUAfcorXe3W2bx4BpXRZNA67SWr+glLIAvwYuAgLAvVrrhwcg/rAYntAY51JTEEIMoZREG/On5vHe9mquWTKOlMTh39El3JrCA8D9WusJwP3Ag9030FrforWeobWeAdwKOIHXQqtvBMYB44H5wI+UUqVnGHvYTE87IDUFIcTQWza7CI/Xz7qPoqN7aq9JQSk1EpgFrAotWgXMUkrlnGa3LwBPaK1docfXAw9prf1a61rgOeDa/ofdNxZf6PJRgiQFIcTQKspJYVJJBms3R0f31HBqCsVAldbaBxD6fSi0/CRKKTvwGeCRLotHARVdHlf2tP9gsEpSEEJE0PLZRdQ3udiya/iPnhpWm0IfXQVUaq23DuSTZmWl9Gs/r89PQsCF37CQk5cVN4Ph5eSkRjqEISdljg/RWOZlWSk8+fZe3tl2mIsXje3z/kNZ5nCSwgGgUCll0Vr7Qo3GBaHlp/J5TqwlQLBmUAJ8GHrcvebQq7q65n5NXOFITiDRcOO1ODh6NHq6hZ2JnJxUamuPRTqMISVljg/RXOYl0wt46q09bNp+iFG54Z/k+1tm0zT69WW618tHWusjwFZgZWjRSmBLqG3gBEqpImAR8ES3VU8DX1JKmaG2iKuAZ/ocbT+0tXtJNNz4rYlDcTghhDilRdPzsdvMYT/XQri9j24DbldK7QJuDz1GKbVaKTWny3a3Ai9qrZ3d9v8LsA/YDWwAfqy13n9GkYep1eUlyXQTsEl7ghAicpIdNhZMzWfDzhqaWt2RDqdHYbUpaK3LgHmnWH5Jt8d397C/D/hyfwI8Ux01hYB9RCQOL4QQnZbNLuLtLVWs23qIyxaURjqcU4r5O5q9Pj9JhhtscvlICBFZhdnJTCnN4K0tVXh9w7N7aswnBZ/fH6opyOUjIUTkLZtTjPOYi827TmqWHRZiPin4fQESpaYghBgmpo3NYmR6Im8O0+k6Yz4p+DxtWIwAAbuMeySEiDzTMDh/dhF7DjZSXt0U6XBOEvNJIeBqBcCwS01BCDE8LDwrnwSbhTXDsHtqzCcFQkkBqSkIIYaJJIeVc8/K41+f1NDUMry6p8Z8Ugi4O5KCI7KBCCFEF8tmF+H1BXhna1WkQzlB7CcFrwcAw2qPcCRCCHFcflYyU0dnsnaYdU+N/aTg8wJgWgZj7D8hhOi/5XOKaGx2s1EfiXQoneIoKQz/GY+EEPFl6pgsRmYkDqsG57hJCoZVagpCiOHFNAyWzS5i76Em9h0aHt1TYz4p4Au2KcjlIyHEcLTwrHwcdgtrNvU0G8HQivmk0Hn5yCqXj4QQw09igpWFZ+XzwSdHaGx29b7DIIv9pOAPXT6SmoIQYphaNrsInz/A21sPRTqUQZmOc3jx+wCwSE0hKvl8XpzOWrze09/gc+SIiT8KJkUfSOGU2Wq1k5GRg0W+FA1ruZlJTBubxVtbqrh0fglWS+S+r8f+J6Xz8lHsFzUWOZ21OBxJJCfnnXZ+bavVxOuNr6TQW5kDgQAtLU04nbVkZ+cPYWSiP5bPLuKXT33Eh2VHmD8lL2JxxP7lI2lTiGper5vk5LTTJgRxaoZhkJyc1mstSwwPk0dnkpeZFPHpOmM+KRihNgWL1BSiliSE/pPXLnp0dE/df7iJvYcaIxdHxI48RAJ+H76AgWnGfFHFEPF6vTz88APccMPV3HrrDXzuc5/hnnvuYvnyhRw7duyEbTdv3sh1111JIBCIULQimiyYmkdiQmRHT435r8+G34sXi3xjEgPmnnvuwuVq55FH/kJSUjJer5eXX36Bmppq1qx5jauuuqZz29WrX+SSSy6Xz58IS7B7agFrNx/k2qXjyEhNGPIYYv/rs9+HLw6KKYbGgQOVrFv3Ft/73n+TlBQcjt1qtXLllVfzqU9dw8svv9i5bWtrC+vWvc3FF18WqXBFFDp/diF+f4C3t0Rm9NSYryng9+ELSFKIBes/Psy72w6fcp1hwJlcoVk4LZ9zz+q9h86uXZqiolGkpaWd/BwLz+O+++5l//59jB49hrVr32Dq1LPIzY1cTxIRfXIzgt1T395axWULSof8+DF9tvQ3VlNY/yH+2C6mGCasVisXXHARq1cHawsvv/wil156RYSjEtFo+ZxijrV6+OCTmiE/dkzXFLwHPgbAbngjHIkYCOee1fO3+aG6T2HCBMXBg5U0NTWdsrZw6aVX8s1vfpVLL72CiopyFi1aMugxidgzuTSD/Kwk3tx0kCuXjh/SY8f0V+iOeZltSFIQA6O4eBTnnruYX/ziHlpbWwDw+Xy8+OJztLa2MnbsOHJycvmf//khK1ZciN0ukzuJvjMMg+VziqmoPsauSueQHjvGk0KoIdCIrztdxeD6r/+6i6KiYj7/+Zu5+ebruPXWG6ioKO9MAJdeegVlZTvl0pE4Iwum5DF34kgcCUN7QceIgv7TpcD+urpm/P6+xeo9rGl78acApP7bnwc8sOEqJyeV2tpjvW8YBaqrK8jLK+l1OxnmomfhvobRIJY+2+Hqb5lN0yArKwVgNFAe7n5hpSCl1ATgUSALqANu0VrvPsV21wH/DRhAAFiuta5RSv0I+ArQMQTgeq31V8MNsr+MhKTBPoQQQsSUcOslDwD3a60fV0rdBDwInN91A6XUHOBHwPla62ql1Aig6+Dgj2mtvz0AMYfNsEtSEEKIvui1TUEpNRKYBawKLVoFzFJK5XTb9D+A+7TW1QBa60atdftABttXRkJyJA8vhBBRJ5yaQjFQpbX2AWitfUqpQ6HltV22mwzsV0qtA1KAvwN3a607GgJuUEpdAFQDP9Rav9+XQEPXxvokEEihOfR3Tk5qn/ePZrFS3iNHTKzW8PpDhLtdLAmnzKZpxsznAWLns90XQ1nmgWzWtgDTgBWAHXgVqAQeI3j56W6ttUcptQJ4Xik1SWtdF+6T96ehuat4apyKpcY4v98fVmOqNDT3zO/3x8znIZY+2+EagIbmvu0XxjYHgEKllAUg9LsgtLyrSuAZrbVLa30MeB44G0BrXa219oT+fiO079Q+R9tPLSQO1aGEECKq9ZoUtNZHgK3AytCilcAWrXVtt03/ClyglDKUUjZgGfARgFKqsGMjpdQMgt1M9RlH34vmNg8/bLyWx+w3DvahhBAiJoR7EfY24Hal1C7g9tBjlFKrQ72OAP4GHAF2EkwiO4A/htbdo5TarpT6CHgIuLmjQXowffBJDQ2+RI55ZdY1MTAWLpxDa2trn/bZvHkjX/jCzadcd/jwIS69dNkZxbR580Y++GDDGT2HEB3CalPQWpcB806x/JIuf/uBb4Z+um936xnE2G+20OTXrS4Z5kLEri1bNtHW1sbZZ58T6VBEDIjpAfGSHMEaQrvbF+FIRCx55pm/sW7d2zQ2NvLVr36dJUuC3/Q3bHiPBx/8LX6/n/T0DL7znTsoKio+af9nn32Kp576K8nJycyfv/CEda+88hKrVv0FwzAoKCjiu9+9g4yMTP74xwdpa2vja1/7BgB//OODuFztXHjhpTz//N/x+/1s3PgBy5ZdwM03f3bQXwMRu2I6KSQ7gsVrl5pCTPDsWo9HrzvlOsMwzmjKS5tajG3CuWFtm5yczMMPP8a2bVu5884fsGTJMpzOev7nf+7kN7/5A6NHj+Gll57jrrv+i4ceevSEfffs2c1jjz3Cn/70BJmZWdx3372d6/bt28MDD/yWP/7xcbKzs3nood/zq1/9gh//+Kc9xjJ27DiuvPLqExKGEGcipjt2JycGawrDfnQnEVWWLbsQgClTzuLo0VpcLhc7dmxn7NgJjB49BoBLLrmCPXt2dY6k2mHLlk0sWLCQzMwsAK688lOd6zZv3sj8+eeSnZ0dWnc1Gzd+MBRFEqJTXNQURGywTTi3x2/zQ3mfQsdoqBaLBQgOnT3YLBYLgcDx8rnd7kE/pohPMV1TSJKkIIbIlClnsXfvLioqyoFg28D48apzHucOM2fO5v331+N01gPw0kvPd66bNWsO77+/nrq6owC8+OJzzJ17NgBFRcVoXYbf76e1tYX33vtn537Jycm0tDQjxECI6bNmgs0S6RBEnMjIyOC//uvH3HXXf+Lz+UhPz+DOO39y0nbjxo3n5ps/x5e//AWSkpKZP/94zWfMmHHcdtvX+I//+GqoobmQ73znDgDOO+981qx5gxtvvIbc3DyUmtS53+LFS7njju/w2c9+RhqaxRmL6fkUAD5/71oAHvn++b1sGTtiaSgAmU+hZzKfQnwY6vkUYvryUYeS3PgbQEsIIfojpi8fATz900tx1rf0vqEQQojYryk47FaslpgvphBCDAg5W4phLwravYYtee1EX0lSEMOa1WqnpaVJTm79EAgEaGlpwmq1RzoUEUVivk1BRLeMjByczlqamxtOu51pmvj98dX7KJwyW612MjK6z5wrRM8kKYhhzWKxkp2d3+t20lVRiIEhl4+EEEJ0kqQghBCiUzRcPrJA8O68/jqTfaOVlDk+SJnjQ3/K3GWfPo33Ew3DXCwE/tnrVkIIIU5lEfBuuBtHQ1JIAOYChwGZQk0IIcJjAfKBDwFXuDtFQ1IQQggxRKShWQghRCdJCkIIITpJUhBCCNFJkoIQQohOkhSEEEJ0kqQghBCikyQFIYQQnaJhmIt+UUpNAB4FsoA64Bat9e7IRnVmlFL3AZ8GSoGztNbbQ8t7LGu0vw5KqSzgL8BYwA3sBv6f1rpWKXUO8CCQSHBi8pu01kdC+/W4LhoopZ4jOOG6H2gGbtdab43l97qDUuqHwI8IfcZj/H0uB9pDPwDf01q/Fskyx3JN4QHgfq31BOB+gi9itHsOWAxUdFt+urJG++sQAH6utVZa67OAvcC9SikTeBz4aqhs64B7AU63LorcqrWerrWeCdwHPBJaHsvvNUqpWcA5hD7jcfA+A1yjtZ4R+nkt0mWOyaSglBoJzAJWhRatAmYppaJ6thGt9bta6wNdl52urLHwOmit67XWb3dZtAEoAWYD7VrrjjFdHgCuC/19unVRQWvd2OXhCMAf6++1UiqBYDL7cpfFMf0+9yCiZY7JpAAUA1Vaax9A6Peh0PJYc7qyxtTrEPqW9GXgBWAUXWpMWuujgKmUyuxlXdRQSj2slKoE7gZuJfbf6x8Dj2uty7ssi/n3GXhCKbVNKfU7pVQ6ES5zrCYFEZt+Q/D6+m8jHchQ0Fp/UWs9CrgD+EWk4xlMSqn5wBzgd5GOZYgt0lpPJzjop8Ew+GzHalI4ABQqpSwAod8FoeWx5nRljZnXIdTIPh64XmvtByoJXkbqWJ8N+LXW9b2sizpa678AS4GDxO57fR4wCdgfanwtAl4DxhHD73PH5WCttYtgQjyXCH+2YzIphFritwIrQ4tWAlu01rWRi2pwnK6ssfI6KKXuIXgt9arQPw/AJiBRKbUw9Pg24Okw1g17SqkUpVRxl8eXA/VAzL7XWut7tdYFWutSrXUpwQR4IcEaUqy+z8lKqRGhvw3gBoLvYUQ/2zE7dLZSaiLB7nkZgJNg9zwd2fFEd4oAAAChSURBVKjOjFLq18DVQB5wFKjTWk85XVmj/XVQSk0BtgO7gLbQ4v1a608ppRYQ7GHj4HjXvJrQfj2uG+6UUrnA80AywTlE6oFva603x/J73VWotnBZqEtqrL7PY4BnCc57YAF2Al/XWh+OZJljNikI8f+3Y8cmAAAgAMP+/9qtu46SPFEosPdyHwFwIwoARBQAiCgAEFEAIKIAQEQBgIgCABksUZ91YOSQ+gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "X_train, X_holdout, y_train, y_holdout = train_test_split(\n",
    "    X, y, test_size=0.3, random_state=17\n",
    ")\n",
    "\n",
    "\n",
    "cv_scores, holdout_scores = [], []\n",
    "n_neighb = [1, 2, 3, 5] + list(range(50, 550, 50))\n",
    "\n",
    "for k in n_neighb:\n",
    "\n",
    "    knn_pipe = Pipeline(\n",
    "        [(\"scaler\", StandardScaler()), (\"knn\", KNeighborsClassifier(n_neighbors=k))]\n",
    "    )\n",
    "    cv_scores.append(np.mean(cross_val_score(knn_pipe, X_train, y_train, cv=5)))\n",
    "    knn_pipe.fit(X_train, y_train)\n",
    "    holdout_scores.append(accuracy_score(y_holdout, knn_pipe.predict(X_holdout)))\n",
    "\n",
    "plt.plot(n_neighb, cv_scores, label=\"CV\")\n",
    "plt.plot(n_neighb, holdout_scores, label=\"holdout\")\n",
    "plt.title(\"Easy task. kNN fails\")\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图表明，即使我们尝试在较广范围内改变 n_neighbors 参数，基于欧几里得距离的 k-NN 在这个问题上依旧表现不佳。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "下面用决策树训练一个模型，看看它在这个任务上的表现如何。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Decision tree. CV: 1.0, holdout: 1.0\n"
     ]
    }
   ],
   "source": [
    "tree = DecisionTreeClassifier(random_state=17, max_depth=1)\n",
    "tree_cv_score = np.mean(cross_val_score(tree, X_train, y_train, cv=5))\n",
    "tree.fit(X_train, y_train)\n",
    "tree_holdout_score = accuracy_score(y_holdout, tree.predict(X_holdout))\n",
    "print(\"Decision tree. CV: {}, holdout: {}\".format(tree_cv_score, tree_holdout_score))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在这一任务中，决策树完美地解决了问题，在交叉验证和留置集上都得到了 100% 的准确率。其实，k-NN 之所以在这个任务上表现不佳并非该方法本身的问题，而是因为使用了欧几里得距离，因为欧几里得距离没能察觉出有一个特征（成比例）比其他所有特征（噪声）更重要。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 决策树和最近邻方法的优势和劣势"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 决策树"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "优势："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 生成容易理解的分类规则，这一属性称为模型的可解释性。例如它生成的规则可能是「如果年龄不满 25 岁，并对摩托车感兴趣，那么就拒绝发放贷款」。\n",
    "- 很容易可视化，即模型本身（树）和特定测试对象的预测（穿过树的路径）可以「被解释」。\n",
    "- 训练和预测的速度快。\n",
    "- 较少的参数数目。\n",
    "- 支持数值和类别特征。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "劣势："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 决策树对输入数据中的噪声非常敏感，这削弱了模型的可解释性。\n",
    "- 决策树构建的边界有其局限性：它由垂直于其中一个坐标轴的超平面组成，在实践中比其他方法的效果要差。\n",
    "- 我们需要通过剪枝、设定叶节点的最小样本数、设定树的最大深度等方法避免过拟合。\n",
    "- 不稳定性，数据的细微变动都会显著改变决策树。这一问题可通过决策树集成方法来处理（以后的实验会介绍）。\n",
    "- 搜索最佳决策树是一个「NP 完全」（NP-Complete）问题。了解什么是 NP-Complete 请点击 [<i class=\"fa fa-external-link-square\" aria-hidden=\"true\"> 这里</i>](https://baike.baidu.com/item/NP-Complete/15961931?fr=aladdin)。实践中使用的一些推断方法，比如基于最大信息增益进行贪婪搜索，并不能保证找到全局最优决策树。\n",
    "- 倘若数据中出现缺失值，将难以创建决策树模型。Friedman 的 CART 算法中大约 50% 的代码是为了处理数据中的缺失值（现在 sklearn 实现了这一算法的改进版本）。\n",
    "- 这一模型只能内插，不能外推（随机森林和树提升方法也是如此）。也就是说，倘若你预测的对象在训练集所设置的特征空间之外，那么决策树就只能做出常数预测。比如，在我们的黄球和蓝球的例子中，这意味着模型将对所有位于 >19 或 <0 的球做出同样的预测。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 最近邻方法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "优势："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 实现简单。\n",
    "- 研究很充分。\n",
    "- 通常而言，在分类、回归、推荐问题中第一个值得尝试的方法就是最近邻方法。\n",
    "- 通过选择恰当的衡量标准或核，它可以适应某一特定问题。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "劣势："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 和其他复合算法相比，这一方法速度较快。但是，现实生活中，用于分类的邻居数目通常较大（100-150），在这一情形下，k-NN 不如决策树快。\n",
    "- 如果数据集有很多变量，很难找到合适的权重，也很难判定哪些特征对分类/回归不重要。\n",
    "- 依赖于对象之间的距离度量，默认选项欧几里得距离常常是不合理的。你可以通过网格搜索参数得到良好的解，但在大型数据集上的耗时很长。\n",
    "- 没有理论来指导我们如何选择邻居数，故而只能进行网格搜索（尽管基本上所有的模型，在对其超参数进行调整时都使用网格搜索的方法）。在邻居数较小的情形下，该方法对离散值很敏感，也就是说，有过拟合的倾向。\n",
    "- 由于「维度的诅咒」，当数据集存在很多特征时它的表现不佳。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 实验总结"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "本次实验我们通过决策树和最近邻方法在几个简单示例上构建了分类模型，并基于交叉验证的方法对模型进行调优，之后对比了决策树和最近邻方法的优劣情况。这两种方法常常作为机器学习模型的基线，熟悉他们的用法是十分必要的。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<i class=\"fa fa-link\" aria-hidden=\"true\"> 相关链接</i>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- [<i class=\"fa fa-external-link-square\" aria-hidden=\"true\"> scikit-learn 文档</i>](http://scikit-learn.org/stable/documentation.html) \n",
    "- [<i class=\"fa fa-external-link-square\" aria-hidden=\"true\"> scikit-learn 指南</i>](https://github.com/amueller/scipy-2017-sklearn)\n",
    "- [<i class=\"fa fa-external-link-square\" aria-hidden=\"true\"> 机器学习优秀课程</i>](https://github.com/diefimov/MTH594_MachineLearning) \n",
    "- [<i class=\"fa fa-external-link-square\" aria-hidden=\"true\"> 适合决策树和 k-NN 的机器学习算法实现</i>](https://github.com/rushter/MLAlgorithms)\n",
    "- [<i class=\"fa fa-external-link-square\" aria-hidden=\"true\"> 了解实验楼《楼+ 机器学习和数据挖掘课程》</i>](https://www.shiyanlou.com/louplus/)"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}