{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 决策树"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 【关键词】树，信息增益"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 决策树的优缺点"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "优点：计算复杂度不高，输出结果易于理解，对中间值的缺失不敏感，可以处理不相关特征数据。既能用于分类，也能用于回归\n",
    "\n",
    "缺点：可能会产生过度匹配问题\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 一、决策树的原理"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### predict()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "【二十个问题的游戏】\n",
    "\n",
    "游戏的规则很简单：参与游戏的一方在脑海里想某个事物，其他参与者向他提问题，只允许提20个问题，问题的答案也只能用对或错回答。问问题的人通过推断分解，逐步缩小待猜测事物的范围。决策树的工作原理与20个问题类似，用户输人一系列数据 ，然后给出游戏的答案。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们经常使用决策树处理分类问题。近来的调查表明决策树也是最经常使用的数据挖掘算法。它之所以如此流行，一个很重要的原因就是使用者基本上不用了解机器学习算法，也不用深究它是如何工作的。"
   ]
  },
  {
   "attachments": {
    "1.PNG": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "如果以前没有接触过决策树，完全不用担心，它的概念非常简单。即使不知道它也可以通过简单的图形了解其工作原理。\n",
    "\n",
    "决策树分类的思想类似于找对象。现想象一个女孩的母亲要给这个女孩介绍男朋友，于是有了下面的对话：\n",
    "\n",
    "      女儿：多大年纪了？\n",
    "\n",
    "      母亲：26。\n",
    "\n",
    "      女儿：长的帅不帅？\n",
    "\n",
    "      母亲：挺帅的。\n",
    "\n",
    "      女儿：收入高不？\n",
    "\n",
    "      母亲：不算很高，中等情况。\n",
    "\n",
    "      女儿：是公务员不？\n",
    "\n",
    "      母亲：是，在税务局上班呢。\n",
    "\n",
    "      女儿：那好，我去见见。\n",
    "      \n",
    "这个女孩的决策过程就是典型的分类树决策。相当于通过年龄、长相、收入和是否公务员对将男人分为两个类别：见和不见。假设这个女孩对男人的要求是：30岁以下、长相中等以上并且是高收入者或中等以上收入的公务员，那么这个可以用下图表示女孩的决策逻辑:\n",
    "\n",
    "![1.PNG](attachment:1.PNG)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上图完整表达了这个女孩决定是否见一个约会对象的策略，其中绿色节点表示判断条件，橙色节点表示决策结果，箭头表示在一个判断条件在不同情况下的决策路径，图中红色箭头表示了上面例子中女孩的决策过程。\n",
    "\n",
    "这幅图基本可以算是一颗决策树，说它“基本可以算”是因为图中的判定条件没有量化，如收入高中低等等，还不能算是严格意义上的决策树，如果将所有条件量化，则就变成真正的决策树了。\n",
    "\n",
    "有了上面直观的认识，我们可以正式定义决策树了：\n",
    "\n",
    "决策树（decision tree）是一个树结构（可以是二叉树或非二叉树）。其每个非叶节点表示一个特征属性上的测试，每个分支代表这个特征属性在某个值域上的输出，而每个叶节点存放一个类别。使用决策树进行决策的过程就是从根节点开始，测试待分类项中相应的特征属性，并按照其值选择输出分支，直到到达叶子节点，将叶子节点存放的类别作为决策结果。\n",
    "\n",
    "可以看到，决策树的决策过程非常直观，容易被人理解。目前决策树已经成功运用于医学、制造产业、天文学、分支生物学以及商业等诸多领域。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "之前介绍的K-近邻算法可以完成很多分类任务，但是它最大的缺点就是无法给出数据的内在含义，决策树的主要优势就在于数据形式非常容易理解。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "决策树算法能够读取数据集合，构建类似于上面的决策树。决策树很多任务都是为了数据中所蕴含的知识信息，因此决策树可以使用不熟悉的数据集合，并从中提取出一系列规则，机器学习算法最终将使用这些机器从数据集中创造的规则。专家系统中经常使用决策树，而且决策树给出结果往往可以匹敌在当前领域具有几十年工作经验的人类专家。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "知道了决策树的定义以及其应用方法，下面介绍决策树的构造算法。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 二、决策树的构造"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 分类解决离散问题，回归解决连续问题"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 决策树：信息论\n",
    "- 逻辑斯底回归、贝叶斯：概率论"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "不同于逻辑斯蒂回归和贝叶斯算法，决策树的构造过程不依赖领域知识，它使用属性选择度量来选择将元组最好地划分成不同的类的属性。所谓决策树的构造就是进行属性选择度量确定各个特征属性之间的拓扑结构。\n",
    "\n",
    "构造决策树的关键步骤是分裂属性。所谓分裂属性就是在某个节点处按照某一特征属性的不同划分构造不同的分支，其目标是让各个分裂子集尽可能地“纯”。尽可能“纯”就是尽量让一个分裂子集中待分类项属于同一类别。分裂属性分为三种不同的情况：\n",
    "\n",
    "      1、属性是离散值且不要求生成二叉决策树。此时用属性的每一个划分作为一个分支。\n",
    "\n",
    "      2、属性是离散值且要求生成二叉决策树。此时使用属性划分的一个子集进行测试，按照“属于此子集”和“不属于此子集”分成两个分支。\n",
    "\n",
    "      3、属性是连续值。此时确定一个值作为分裂点split_point，按照>split_point和<=split_point生成两个分支。\n",
    "\n",
    "构造决策树的关键性内容是进行属性选择度量，属性选择度量是一种选择分裂准则，它决定了拓扑结构及分裂点split_point的选择。\n",
    "\n",
    "属性选择度量算法有很多，一般使用自顶向下递归分治法，并采用不回溯的贪心策略。这里介绍常用的ID3算法。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ID3算法"
   ]
  },
  {
   "attachments": {
    "10.gif": {
     "image/gif": "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"
    },
    "11.png": {
     "image/png": "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"
    },
    "2.PNG": {
     "image/png": "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"
    },
    "3.PNG": {
     "image/png": "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"
    },
    "4.gif": {
     "image/gif": "R0lGODlh0AAzALMAAP///wAAAKqqqhAQEO7u7tzc3Lq6upiYmGZmZszMzHZ2diIiIoiIiFRUVDIyMkRERCH5BAEAAAAALAAAAADQADMAAAT+EMhJq7046827zwSjGEdTHAfyrWzrvnAsz5MABAnAPNJA/8CgcEicJHiAho2wKDqf0KiQwZAsCICUdMvteimPHFPSIH3P6PQMaVBIDgyDek6v2+/4vH7P7/v/gIGCg4SFhoeIiX8DAQEDj5CRj42UjTaKmGcHjUgbJw2NDpmjXg+NVR4ECjikZwQHLAdYagSMrB8CKnivK7JouisEwGkJjU0rCrMtBahBwx3CFgI+LMIMzxgMORUKCwEK4AgPchMGzWmrAQ1SCgOwQdot5hUEBS0PArkdohcP6xNM3kngN8eBJSkL7AUhyILhCwIBlG1AgSGAQAkGIk5gcImWLYX+TmoJoeiCowwBxzo8AEkho0QAEMkBSPBvTsYADocIqPlj5QuaGLVIEEDlVQmZElQtcDCiAtEDCrYBoGahm4UCFilQnYOgEbYORBmcEGqhwUUAT6NSUJUCwbkKW9FSGft1gg9eA8QYwPpOQE4AD84CcJuUatwJDrBlRBrgjrcASD0IgPiOsIUBLC2LlFDgCkzIGeJOzjr4LYXGcrBiRJtyx4W8FQ4wBN25YscJqyR6toCgge/fwN0Mwero5YYEBBJQVfBPgPACW2WfJudAoEaiCoDVNpJ8eXPhE3Znp4AAfAPTEC8LhJhjewWXFhbwBLB7xpEH+PPnN31hU4BOHzD+UNMDuigjoFbrsQIfAAY0EQIZz6W00YAFxjcLbHbJFABLGEkIAFYgCdAYTIcBYFVsGmmVByi3eRDYBKSBIRCINYzIQErjGcCPdCRa8KIEMV42VBPkqDYkgxUcSIGRAzVT4gLgcTZAZFNh0BtwWEYphAP8cZAiShIoeUMB9Xw4YpMSMIfYJQqN14MFXx4jpl1puoEKj4PZWcF58Sk0p09rgVbOAlQCdYdvLhSjkAM5wOJZegIlFGZNAiCxCYeSSgCoBIoO1Ch9EhmaAAK+mChQAgow8BKhV5HKwFkkpWlQA+HweYFJdijwVwcloKAASAUQpMABHRXg6lmpEhVXGRT+xJqFWVAB6xCuL5yZCoAN3XFAfSyYdcGwMai5UXsAAehtVWftygEWBmDbgVryCIaGAZjBgGEFV8jrQXU1TMYMBfBOJRUF+ZajLwfuNNVCXRpEUwd0VDasDJPciIUMP35NQFwjWupCMcAWw8QwB8N2ue7BGJQ6R0AsAKNjAAgY54Jz2H3wyssxTyRzK0UwFct8PAcdBKIfFABKxEInDUM3OydVQALOGeSI0lTPIGIlWGeNtZZVdx0gOGCHLfbYyXht9tlop6322my3DYNzHBTgrttuNz0BVB7SrXfcee+t9pwZuOf32o4SIAyWvikk+OBqB9vB4oyjDS7fkbOdLwEUiCfOWd+Vd50qhxYY0FXInZeedAQAOw=="
    },
    "5.gif": {
     "image/gif": "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"
    },
    "6.gif": {
     "image/gif": "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"
    },
    "7.png": {
     "image/png": "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"
    },
    "8.gif": {
     "image/gif": "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"
    },
    "9.gif": {
     "image/gif": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "划分数据集的大原则是：*将无序的数据变得更加有序。*\n",
    "\n",
    "我们可以使用多种方法划分数据集，但是每种方法都有各自的优缺点。组织杂乱无章数据的一种方法就是使用信息论度量信息，信息论是量化处理信息的分支科学。我们可以在划分数据之前使用信息论量化度量信息的内容。\n",
    "\n",
    "在划分数据集之前之后信息发生的变化称为信息增益，知道如何计算信息增益，我们就可以计算每个特征值划分数据集获得的信息增益，获得信息增益最高的特征就是最好的选择。\n",
    "\n",
    "在可以评测哪种数据划分方式是最好的数据划分之前，我们必须学习如何计算信息增益。集合信息的度量方式称为香农熵或者简称为熵，这个名字来源于信息论之父克劳德•香农。\n",
    "\n",
    "entropy\n",
    "\n",
    "熵定义为信息的期望值，在明晰这个概念之前，我们必须知道信息的定义。如果待分类的事务可能划分在多个分类之中，则符号x的信息定义为：\n",
    "\n",
    "![2.PNG](attachment:2.PNG)\n",
    "\n",
    "其中p(x)是选择该分类的概率\n",
    "\n",
    "为了计算熵，我们需要计算所有类别所有可能值包含的信息期望值，通过下面的公式得到：\n",
    "\n",
    "![3.PNG](attachment:3.PNG)\n",
    "\n",
    "其中n是分类的数目。\n",
    "\n",
    "在决策树当中，设D为用类别对训练元组进行的划分，则D的熵（entropy）表示为：\n",
    "\n",
    "![4.gif](attachment:4.gif)\n",
    "\n",
    "其中pi表示第i个类别在整个训练元组中出现的概率，可以用属于此类别元素的数量除以训练元组元素总数量作为估计。熵的实际意义表示是D中元组的类标号所需要的平均信息量。\n",
    "\n",
    "现在我们假设将训练元组D按属性A进行划分，则A对D划分的期望信息为：\n",
    "\n",
    "![5.gif](attachment:5.gif)\n",
    "\n",
    "而信息增益即为两者的差值：\n",
    "\n",
    "![6.gif](attachment:6.gif)\n",
    "\n",
    "ID3算法就是在每次需要分裂时，计算每个属性的增益率，然后选择增益率最大的属性进行分裂。下面我们继续用SNS社区中不真实账号检测的例子说明如何使用ID3算法构造决策树。为了简单起见，我们假设训练集合包含10个元素：\n",
    "\n",
    "![7.png](attachment:7.png)\n",
    "\n",
    "其中s、m和l分别表示小、中和大。\n",
    "\n",
    "设L、F和H表示日志密度、好友密度、是否使用真实头像，下面计算各属性的信息增益。\n",
    "\n",
    "![8.gif](attachment:8.gif)\n",
    "![9.gif](attachment:9.gif)\n",
    "![10.gif](attachment:10.gif)\n",
    "\n",
    "因此日志密度的信息增益是0.276。\n",
    "\n",
    "用同样方法得到F和H的信息增益分别为0.553和0.033。\n",
    "\n",
    "因为F具有最大的信息增益，所以第一次分裂选择F为分裂属性，分裂后的结果如下图表示：\n",
    "\n",
    "![11.png](attachment:11.png)\n",
    "\n",
    "在上图的基础上，再递归使用这个方法计算子节点的分裂属性，最终就可以得到整个决策树。\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 练习\n",
    "计算上图的信息熵，确定下一个分类的特征"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import math as math"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8812908992306927"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#求解账号是否真实的熵\n",
    "#no 0.3 yes 0.7\n",
    "\n",
    "info_D = -(0.3*math.log2(0.3)+0.7*math.log2(0.7))\n",
    "\n",
    "info_D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#绘制二叉树：日志密度，好友密度，是否真实头像\n",
    "\n",
    "# s 0.3 m 0.4 l 0.3\n",
    "\n",
    "#s ----> 2/3 no  1/3 yes\n",
    "# m-----> 1/4 no  3/4 yes\n",
    "# l----> 0 no  1 yes\n",
    "info_L_D = -(0.3*(2/3*math.log2(2/3)+1/3*math.log2(1/3))  \n",
    "             + 0.4*(1/4*math.log2(1/4)+3/4*math.log2(3/4)) \n",
    "             + 0.3*(1*math.log2(1)))\n",
    "info_L_D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.2812908992306927"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#信息增益\n",
    "#按照日志密度进行划分\n",
    "info_D - info_L_D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.32451124978365314"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#好友密度进行划分，求解信息熵\n",
    "\n",
    "# s 0.4  m 0.4  l 0.2\n",
    "\n",
    "\n",
    "# s 3/4 no 1/4 yes\n",
    "\n",
    "# m 1 yes\n",
    "\n",
    "# l yes 1\n",
    "info_F_D = -(0.4*(3/4*math.log2(3/4)+1/4*math.log2(1/4)) \n",
    "             + 0.4*(1*math.log2(1)) \n",
    "             + 0.2*(1*math.log2(1)))\n",
    "info_F_D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5567796494470396"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#求解以好友密度进行划分的信息增益\n",
    "info_D - info_F_D\n",
    "\n",
    "# 跟刚才的按照日志密度进行划分相比，按照好友密度进行划分，信息增益大，所以选择按照好友密度进行划分"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 三、实战"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "【注意】\n",
    "参数max_depth越大，越容易过拟合"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "import sklearn.datasets as datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1、使用自带的iris数据集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "iris = datasets.load_iris()\n",
    "\n",
    "x_data = iris.data\n",
    "y_target = iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train,x_test,y_train,y_test = train_test_split(x_data,y_target,test_size = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用决策树算法"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#max_depth 不进行声明，那么，有多少个属性，树的深度就是多少\n",
    "#如果属性太多，此时就需要限定树的深度\n",
    "#max_depth 最大的深度，属性200，max_depth 根据信息增益，进行选择，最重要的100个属性，然后进行数据的分类\n",
    "tree = DecisionTreeClassifier(max_depth=5)\n",
    "\n",
    "# tree.fit(X_train,y_train)\n",
    "\n",
    "tree.fit(x_data,y_target).score(x_data,y_target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.73333333333333328"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree.score(x_test,y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用KNN算法"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.96666666666666667"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "knn = KNeighborsClassifier(5)\n",
    "\n",
    "knn.fit(x_data,y_target).score(x_data,y_target)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用逻辑斯蒂回归算法"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.95999999999999996"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 决策树 ：信息论 信息熵将划分展示\n",
    "# 逻辑斯底：概率论\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "lrg = LogisticRegression()\n",
    "\n",
    "lrg.fit(x_data,y_target).score(x_data,y_target)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2、使用回归预测一个椭圆"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用RandomState生成固定随机数  \n",
    "创建-100到100之间的角度  \n",
    "生成正弦值和余弦值  \n",
    "添加噪声"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.tree import DecisionTreeRegressor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-98.25897143],\n",
       "       [-97.61353655],\n",
       "       [-93.7845092 ],\n",
       "       [-92.638643  ],\n",
       "       [-88.32920567],\n",
       "       [-87.74078938],\n",
       "       [-85.36860635],\n",
       "       [-85.24688731],\n",
       "       [-83.34651117],\n",
       "       [-80.38829665],\n",
       "       [-80.1209589 ],\n",
       "       [-76.71435762],\n",
       "       [-76.40968096],\n",
       "       [-72.09652444],\n",
       "       [-70.85309375],\n",
       "       [-70.01671507],\n",
       "       [-68.23069059],\n",
       "       [-68.17500301],\n",
       "       [-67.26146831],\n",
       "       [-66.14433432],\n",
       "       [-64.77602684],\n",
       "       [-63.63303187],\n",
       "       [-61.12805278],\n",
       "       [-60.57529221],\n",
       "       [-57.99092137],\n",
       "       [-51.99071746],\n",
       "       [-50.19521255],\n",
       "       [-44.9918581 ],\n",
       "       [-44.71462095],\n",
       "       [-39.84789847],\n",
       "       [-37.44971539],\n",
       "       [-36.8924444 ],\n",
       "       [-36.78921826],\n",
       "       [-34.68277363],\n",
       "       [-28.20571676],\n",
       "       [-25.55771831],\n",
       "       [-24.02080064],\n",
       "       [-21.71298869],\n",
       "       [-20.88656494],\n",
       "       [-19.59548861],\n",
       "       [-17.61823774],\n",
       "       [-15.95916267],\n",
       "       [-13.57216989],\n",
       "       [-13.38717073],\n",
       "       [-12.84505049],\n",
       "       [-11.22206503],\n",
       "       [-10.81969008],\n",
       "       [ -9.30435017],\n",
       "       [ -9.21530701],\n",
       "       [ -8.5172761 ],\n",
       "       [ -5.52538632],\n",
       "       [ -4.34231591],\n",
       "       [ -0.69508746],\n",
       "       [  9.8174906 ],\n",
       "       [ 10.77651454],\n",
       "       [ 12.5120284 ],\n",
       "       [ 17.91758471],\n",
       "       [ 19.00066293],\n",
       "       [ 20.59787053],\n",
       "       [ 21.78355563],\n",
       "       [ 26.58743695],\n",
       "       [ 33.43388355],\n",
       "       [ 34.36501088],\n",
       "       [ 34.8101705 ],\n",
       "       [ 36.77783926],\n",
       "       [ 36.80159782],\n",
       "       [ 36.98169858],\n",
       "       [ 39.02306412],\n",
       "       [ 39.96763632],\n",
       "       [ 40.69821854],\n",
       "       [ 41.55507059],\n",
       "       [ 47.71671998],\n",
       "       [ 51.27621405],\n",
       "       [ 55.58728132],\n",
       "       [ 57.27564337],\n",
       "       [ 57.96236177],\n",
       "       [ 60.7774663 ],\n",
       "       [ 61.67580216],\n",
       "       [ 63.05125616],\n",
       "       [ 67.07730481],\n",
       "       [ 68.4992789 ],\n",
       "       [ 69.44889837],\n",
       "       [ 71.78064256],\n",
       "       [ 71.92220066],\n",
       "       [ 72.4169162 ],\n",
       "       [ 74.15564907],\n",
       "       [ 75.85083296],\n",
       "       [ 81.39358055],\n",
       "       [ 81.91868744],\n",
       "       [ 82.01406723],\n",
       "       [ 85.05218396],\n",
       "       [ 87.549674  ],\n",
       "       [ 88.95254518],\n",
       "       [ 89.55029014],\n",
       "       [ 90.0691102 ],\n",
       "       [ 91.40841773],\n",
       "       [ 93.06252197],\n",
       "       [ 93.66440172],\n",
       "       [ 94.27864841],\n",
       "       [ 99.13917816]])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#-100 ~ 100 无序\n",
    "# np.sort() 排序\n",
    "# 原始数据是二维的，所以，axis指定排序的维度\n",
    "#sort 聚合操作\n",
    "x_data = np.sort(200*np.random.rand(100,1) - 100,axis = 0)\n",
    "x_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#使用随机创造的数据，生成圆上的点\n",
    "\n",
    "dot_x = np.pi*np.sin(x_data)\n",
    "\n",
    "dot_y = np.pi*np.cos(x_data)\n",
    "\n",
    "y_target = np.c_[dot_x.ravel(),dot_y.ravel()]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 2.40039799, -2.02674466],\n",
       "       [ 0.69834979, -3.06299069],\n",
       "       [ 1.40390294,  2.81045565],\n",
       "       [ 3.1392839 , -0.12042003],\n",
       "       [-1.12024869,  2.93507194],\n",
       "       [ 0.69724898,  3.06324146],\n",
       "       [ 1.63028268, -2.68547627],\n",
       "       [ 1.29215378, -2.86355426],\n",
       "       [-3.12763298, -0.2958316 ],\n",
       "       [ 3.02124737,  0.86120193],\n",
       "       [ 3.14142451,  0.03250308],\n",
       "       [-3.04027126,  0.79142596],\n",
       "       [-2.66283282,  1.66701102],\n",
       "       [-0.50084351, -3.10141261],\n",
       "       [-3.09775105, -0.52301321],\n",
       "       [-2.46416492,  1.94871641],\n",
       "       [ 2.43002753,  1.99112296],\n",
       "       [ 2.53708412,  1.852784  ],\n",
       "       [ 3.01682494, -0.87656812],\n",
       "       [ 0.53425316, -3.09583235],\n",
       "       [-2.9251391 , -1.14593439],\n",
       "       [-2.25621917,  2.18611057],\n",
       "       [ 3.11384612, -0.41661341],\n",
       "       [ 2.43138888, -1.98946035],\n",
       "       [-3.11567374,  0.40271768],\n",
       "       [-3.10420152, -0.48325701],\n",
       "       [ 0.2205778 ,  3.13383947],\n",
       "       [-2.65966684,  1.67205762],\n",
       "       [-2.10046718,  2.33615968],\n",
       "       [-2.63127963, -1.71638337],\n",
       "       [ 0.77540516,  3.0443967 ],\n",
       "       [ 2.26818391,  2.17369412],\n",
       "       [ 2.48009394,  1.92840308],\n",
       "       [ 0.39247032, -3.11698114],\n",
       "       [-0.21539794, -3.13419976],\n",
       "       [-1.29527834,  2.86214228],\n",
       "       [ 2.81662601,  1.39148207],\n",
       "       [-0.86263966, -3.02083717],\n",
       "       [-2.8063128 , -1.41216602],\n",
       "       [-2.13206423,  2.30735921],\n",
       "       [ 2.96229747,  1.04613485],\n",
       "       [ 0.78089279, -3.04299373],\n",
       "       [-2.65335831,  1.68205056],\n",
       "       [-2.29867648,  2.14142262],\n",
       "       [-0.86421034,  3.0203882 ],\n",
       "       [ 3.06135773,  0.70547379],\n",
       "       [ 3.09312481, -0.54971203],\n",
       "       [-0.37742122, -3.11883915],\n",
       "       [-0.65327044, -3.07292078],\n",
       "       [-2.47547465, -1.93432925],\n",
       "       [ 2.15929329,  2.28189765],\n",
       "       [ 2.92890972, -1.13626239],\n",
       "       [-2.01204121,  2.41273591],\n",
       "       [-1.20227481, -2.90243685],\n",
       "       [-3.06651553, -0.68270558],\n",
       "       [-0.17063709,  3.13695511],\n",
       "       [-2.52206066,  1.87318297],\n",
       "       [ 0.47291216,  3.10579434],\n",
       "       [ 3.09222241, -0.55476568],\n",
       "       [ 0.64749834, -3.07414221],\n",
       "       [ 3.12044311,  0.36392195],\n",
       "       [ 2.83270647, -1.35844708],\n",
       "       [ 0.60105412, -3.08355936],\n",
       "       [-0.78531012, -3.04185674],\n",
       "       [-2.50187621,  1.90005784],\n",
       "       [-2.45603173,  1.958957  ],\n",
       "       [-2.06540159,  2.36721791],\n",
       "       [ 3.04636592,  0.76763211],\n",
       "       [ 2.40741432, -2.01840548],\n",
       "       [ 0.44611971, -3.10975588],\n",
       "       [-2.0581787 , -2.37350055],\n",
       "       [-1.7552408 , -2.60551993],\n",
       "       [ 2.6616224 ,  1.66894296],\n",
       "       [-2.57606179,  1.79819634],\n",
       "       [ 2.08794297,  2.34735991],\n",
       "       [ 3.10290328,  0.49152378],\n",
       "       [-2.78135005, -1.46071774],\n",
       "       [-2.87524509,  1.26592656],\n",
       "       [ 0.68375841,  3.06628095],\n",
       "       [-2.80528889, -1.41419894],\n",
       "       [-1.81451694,  2.5645921 ],\n",
       "       [ 1.02947578,  2.96812803],\n",
       "       [ 1.43953199, -2.79237391],\n",
       "       [ 1.0311686 , -2.96754035],\n",
       "       [-0.50139733, -3.10132312],\n",
       "       [-2.97388529,  1.01272438],\n",
       "       [ 1.37386107,  2.82526285],\n",
       "       [-0.89180601,  3.01235563],\n",
       "       [ 0.73845712,  3.0535693 ],\n",
       "       [ 1.02590807,  2.96936307],\n",
       "       [-0.71371108, -3.05944781],\n",
       "       [-1.2664294 ,  2.87502365],\n",
       "       [ 2.62291537,  1.72913834],\n",
       "       [ 3.14124395, -0.04680645],\n",
       "       [ 2.70466227, -1.59825105],\n",
       "       [-0.93509697, -2.99919957],\n",
       "       [-2.91098715,  1.18142211],\n",
       "       [-1.73053462,  2.62199434],\n",
       "       [ 0.09696181,  3.14009599],\n",
       "       [-3.09139166,  0.55937645]])"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#扰乱数据\n",
    "#添加噪声\n",
    "y_target[::5] += np.random.randn(20,2)*0.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fa170c83f60>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAD8CAYAAABjAo9vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa170ca52e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(y_target[:,0],y_target[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 2.40039799,  0.69834979,  1.40390294,  3.1392839 , -1.12024869,\n",
       "        0.69724898,  1.63028268,  1.29215378, -3.12763298,  3.02124737,\n",
       "        3.14142451, -3.04027126, -2.66283282, -0.50084351, -3.09775105,\n",
       "       -2.46416492,  2.43002753,  2.53708412,  3.01682494,  0.53425316,\n",
       "       -2.9251391 , -2.25621917,  3.11384612,  2.43138888, -3.11567374,\n",
       "       -3.10420152,  0.2205778 , -2.65966684, -2.10046718, -2.63127963,\n",
       "        0.77540516,  2.26818391,  2.48009394,  0.39247032, -0.21539794,\n",
       "       -1.29527834,  2.81662601, -0.86263966, -2.8063128 , -2.13206423,\n",
       "        2.96229747,  0.78089279, -2.65335831, -2.29867648, -0.86421034,\n",
       "        3.06135773,  3.09312481, -0.37742122, -0.65327044, -2.47547465,\n",
       "        2.15929329,  2.92890972, -2.01204121, -1.20227481, -3.06651553,\n",
       "       -0.17063709, -2.52206066,  0.47291216,  3.09222241,  0.64749834,\n",
       "        3.12044311,  2.83270647,  0.60105412, -0.78531012, -2.50187621,\n",
       "       -2.45603173, -2.06540159,  3.04636592,  2.40741432,  0.44611971,\n",
       "       -2.0581787 , -1.7552408 ,  2.6616224 , -2.57606179,  2.08794297,\n",
       "        3.10290328, -2.78135005, -2.87524509,  0.68375841, -2.80528889,\n",
       "       -1.81451694,  1.02947578,  1.43953199,  1.0311686 , -0.50139733,\n",
       "       -2.97388529,  1.37386107, -0.89180601,  0.73845712,  1.02590807,\n",
       "       -0.71371108, -1.2664294 ,  2.62291537,  3.14124395,  2.70466227,\n",
       "       -0.93509697, -2.91098715, -1.73053462,  0.09696181, -3.09139166])"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dot_x.ravel()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "创建不同深度的决策树  \n",
    "进行数据训练"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tree_2 = DecisionTreeRegressor(max_depth=2)\n",
    "tree_5 = DecisionTreeRegressor(max_depth=5)\n",
    "tree_20 = DecisionTreeRegressor(max_depth=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DecisionTreeRegressor(criterion='mse', max_depth=20, max_features=None,\n",
       "           max_leaf_nodes=None, min_impurity_split=1e-07,\n",
       "           min_samples_leaf=1, min_samples_split=2,\n",
       "           min_weight_fraction_leaf=0.0, presort=False, random_state=None,\n",
       "           splitter='best')"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_2.fit(x_data,y_target)\n",
    "tree_5.fit(x_data,y_target)\n",
    "tree_20.fit(x_data,y_target)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "创建-100到100的预测数据，间隔为0.01  \n",
    "对数据进行预测"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20000,)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#创造 预测数据\n",
    "x_test = np.arange(-100,100,0.01).reshape((20000,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fa170c3e630>"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAD8CAYAAABjAo9vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngGH+HxVX6Pd7B/YkAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa170c4e940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y2_ = tree_2.predict(x_test)\n",
    "\n",
    "plt.scatter(y2_[:,0],y2_[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fa170b643c8>"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAD8CAYAAABjAo9vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa170c4e4e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y5_ = tree_5.predict(x_test)\n",
    "plt.scatter(y5_[:,0],y5_[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fa170c5ccf8>"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAD8CAYAAABjAo9vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa170c3ff98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y20_ = tree_20.predict(x_test)\n",
    "plt.scatter(y20_[:,0],y20_[:,1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "显示图片"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 四、作业"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1、预测隐形眼镜的类型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "分析lenses.txt文件"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sklearn.datasets as datasets\n",
    "import pandas as pd\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>young\tmyope\tno\treduced\tno lenses</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>young\\tmyope\\tno\\tnormal\\tsoft</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>young\\tmyope\\tyes\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>young\\tmyope\\tyes\\tnormal\\thard</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>young\\thyper\\tno\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>young\\thyper\\tno\\tnormal\\tsoft</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>young\\thyper\\tyes\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>young\\thyper\\tyes\\tnormal\\thard</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>pre\\tmyope\\tno\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>pre\\tmyope\\tno\\tnormal\\tsoft</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>pre\\tmyope\\tyes\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>pre\\tmyope\\tyes\\tnormal\\thard</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>pre\\thyper\\tno\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>pre\\thyper\\tno\\tnormal\\tsoft</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>pre\\thyper\\tyes\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>pre\\thyper\\tyes\\tnormal\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>presbyopic\\tmyope\\tno\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>presbyopic\\tmyope\\tno\\tnormal\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>presbyopic\\tmyope\\tyes\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>presbyopic\\tmyope\\tyes\\tnormal\\thard</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>presbyopic\\thyper\\tno\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>presbyopic\\thyper\\tno\\tnormal\\tsoft</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>presbyopic\\thyper\\tyes\\treduced\\tno lenses</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>presbyopic\\thyper\\tyes\\tnormal\\tno lenses</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          young\\tmyope\\tno\\treduced\\tno lenses\n",
       "0               young\\tmyope\\tno\\tnormal\\tsoft\n",
       "1        young\\tmyope\\tyes\\treduced\\tno lenses\n",
       "2              young\\tmyope\\tyes\\tnormal\\thard\n",
       "3         young\\thyper\\tno\\treduced\\tno lenses\n",
       "4               young\\thyper\\tno\\tnormal\\tsoft\n",
       "5        young\\thyper\\tyes\\treduced\\tno lenses\n",
       "6              young\\thyper\\tyes\\tnormal\\thard\n",
       "7           pre\\tmyope\\tno\\treduced\\tno lenses\n",
       "8                 pre\\tmyope\\tno\\tnormal\\tsoft\n",
       "9          pre\\tmyope\\tyes\\treduced\\tno lenses\n",
       "10               pre\\tmyope\\tyes\\tnormal\\thard\n",
       "11          pre\\thyper\\tno\\treduced\\tno lenses\n",
       "12                pre\\thyper\\tno\\tnormal\\tsoft\n",
       "13         pre\\thyper\\tyes\\treduced\\tno lenses\n",
       "14          pre\\thyper\\tyes\\tnormal\\tno lenses\n",
       "15   presbyopic\\tmyope\\tno\\treduced\\tno lenses\n",
       "16    presbyopic\\tmyope\\tno\\tnormal\\tno lenses\n",
       "17  presbyopic\\tmyope\\tyes\\treduced\\tno lenses\n",
       "18        presbyopic\\tmyope\\tyes\\tnormal\\thard\n",
       "19   presbyopic\\thyper\\tno\\treduced\\tno lenses\n",
       "20         presbyopic\\thyper\\tno\\tnormal\\tsoft\n",
       "21  presbyopic\\thyper\\tyes\\treduced\\tno lenses\n",
       "22   presbyopic\\thyper\\tyes\\tnormal\\tno lenses"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len = pd.read_csv('../data/lenses.txt')\n",
    "len"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2、使用make_blobs产生数据，训练模型，并画出类别边界。"
   ]
  }
 ],
 "metadata": {
  "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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}