{ "metadata": { "name": "week8_tutorial" }, "nbformat": 3, "nbformat_minor": 0, "worksheets": [ { "cells": [ { "cell_type": "heading", "level": 1, "metadata": {}, "source": [ "Psych 216A Week 8 Tutorial: Classification" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This tutorial will cover classification based on logistic regression models" ] }, { "cell_type": "code", "collapsed": false, "input": [ "%pylab inline" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ "\n", "Welcome to pylab, a matplotlib-based Python environment [backend: module://IPython.zmq.pylab.backend_inline].\n", "For more information, type 'help(pylab)'.\n" ] } ], "prompt_number": 1 }, { "cell_type": "code", "collapsed": false, "input": [ "import moss\n", "import seaborn\n", "seaborn.set()\n", "colors = seaborn.color_palette()" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 2 }, { "cell_type": "code", "collapsed": false, "input": [ "# Generate two random predictors\n", "X = randn(100, 2)\n", "# Create y as a weighted sum of the 2 predictors and noise\n", "noise = randn(100)\n", "y = 0.7 * X[:, 0] + 0.5 * X[:, 1] + noise\n", "# Threshold y so that it is a binary classification. Let's pretend this is\n", "# medical data and 0 means the patient died and 1 means the patient lived.\n", "# x1 and x2 are measures of symptoms\n", "y = (y > 0).astype(int)" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 3 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Plot the symptoms x1 and x2 for the patients who lived and died. Notice\n", "that outcomes can not be perfectly classified based on x1 and x2. This\n", "is due to the noise we added. Without noise we would could classify\n", "perfectly" ] }, { "cell_type": "code", "collapsed": false, "input": [ "plot(X[y == 0, 0], X[y == 0, 1], \"o\", label=\"dead\")\n", "plot(X[y == 1, 0], X[y == 1, 1], \"o\", color=colors[2], label=\"alive\")\n", "legend();" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAFRCAYAAAA1uqfwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nec4AAAAASUVORK5CYII=\n" } ], "prompt_number": 4 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now we can use a logistic regression to try to classify patients who die\n", "or live based on their symptoms. We will use the scikit-learn package,\n", "which provides a nice interface for logistic regression (and many other\n", "statistical learning models)." ] }, { "cell_type": "code", "collapsed": false, "input": [ "from sklearn.linear_model import LogisticRegression\n", "model = LogisticRegression().fit(X, y)\n", "\n", "# We can call the predict() method of the model to get estimated values\n", "y_hat = model.predict(X)\n", "\n", "# To be closer to the matlab tutorial, you can also use the predict_proba() method\n", "# to return probabilistc estimates in favor of each class\n", "y_hat2 = argmax(model.predict_proba(X), axis=1)\n", "\n", "assert all(y_hat == y_hat2)" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 5 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Make a figure showing the correct and incorrect classifications" ] }, { "cell_type": "code", "collapsed": false, "input": [ "a_c = logical_and(y == 1, y_hat == 1)\n", "a_i = logical_and(y == 1, y_hat == 0)\n", "d_c = logical_and(y == 0, y_hat == 0)\n", "d_i = logical_and(y == 0, y_hat == 1)\n", "plot(X[d_c, 0], X[d_c, 1], \"o\", color=colors[0], label=\"dead +\")\n", "plot(X[d_i, 0], X[d_i, 1], \"o\", color=colors[0], alpha=0.7, label=\"dead -\")\n", "plot(X[a_c, 0], X[a_c, 1], \"o\", color=colors[2], label=\"alive +\")\n", "plot(X[a_i, 0], X[a_i, 1], \"o\", color=colors[2], alpha=0.7, label=\"alive -\")\n", "legend();" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAFRCAYAAAA1uqfwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XlclOXeP/DPLKBohDuQS5mpKCLihlsgCLiiIZrwqHFc\nUs9JycfS6un0q9PvnBYtzeVXT2VuR5RUtAyNTRSe7JHMBTmYnexkamEBFsdIhVl+fyCjyAwzzNxz\nb/N5v169XnLNcM91Xw3znWv7Xhqz2WwGERERiUYrdQWIiIg8DYMvERGRyBh8iYiIRMbgS0REJDIG\nXyIiIpEx+BIREYnMpeB748YNhIeHY8CAARg2bBjWrFkjVL2IiIhUS+PqPt/ff/8drVq1ws2bNzFo\n0CB89NFHeOihh4SqHxERkeq4POzcqlUrAMBvv/0Gg8GAFi1auFwpIiIiNXM5+JpMJoSGhsLf3x+L\nFy9G165dhagXERGRark87FzvwoULmDBhAtLS0hAWFmYpN5nMMBiMQryEaun1OgCQVTsdycnBnpWv\nI6VtB0vZ1l8qMG3FMxgdFydZveTYVnLkKe0kxPvUU9rKVWwnx+j1Omi1GrvPEyz4AsDTTz+Nhx56\nCIsWLbKU1dQYUFV1XaiXUCU/Px8AkFU7vbhgDmaaGr810vRa/OW/N0lQozpybCs58pR2EuJ96ilt\n5Sq2k2P8/Hzg7a23+zyXhp0rKirw66+/AgAqKyuRk5ODKVOmuHJJkgmtyWS9nN96SUb4PiWlsh+e\nm1BWVoaUlBQYjUYEBATg6aefRmBgoFB1IwmZtFrASo/CdGvoiUgO+D4lpXKp5xsSEoKTJ0+iuLgY\n2dnZeOyxx4SqF0ksLmkWNleWNyjbVFGOuEdnSlQjosb4PiWlcqnnS+o1MnoMACBtVxq0BiNMeh3i\nU5dZyonkgO9T+4xGIy5c+M7l6/j61m0jvXbtpsvXUosHHugOnc65URZBF1xZwwVX9nEhg+PYVo5h\nOzlO7W317bfncfHiBXTr9oDUVVGV+jbt0aNhUilHF1yx50tEpHLWggRJiwcrEBERiYzBl4iISGQM\nvkREJBtLlizCa6/9VfDrXrz4Pfz9/WCysTdcbAy+REQkGxqNBhqN/fSMSsfgS0REsuLmTTh2rVr1\nKlatetWtr8HVzkREHiy/oADpGZkwmjXQacxISpyE6MhI0a5x6dJFvPXWG8jKOoixY8fDaDQ0eLy0\n9B/YsmUj8vJyMHnyI3j88T+iS5e60/PWrVuN7du3oarqF0RHxyIlZS6GDRsBoC6A7927G2vWrIKX\nlzf++MfFzbond2PPl4jIQ+UXFGDdxl0w3Dce5s7jYLhvPNZt3IX8ggLRrjF37mz4+vqisLAIPXr0\nxMcf77MMO1+9ehUJCRMRHR2LwsJjaNeuPRYunGv53e7dH0RmZjZOnz6H0NAwLFo0z/JYbm42Xnvt\nb3jzzXXYsOFdbN78vqyGsxl8iYg8VHpGJtoFJzYoaxeciA/3HhDlGj///DO+/vorPPvsC2jfvj2e\neCIVHTt2sjx+4MB+xMdPwfjxE+Hrey8WL16KCxf+hZ9//hkAEB//CDp18oePjw8WLvwTNBoNzpw5\nDQDIy8vBtGmPIjx8OIKD+2H27DmSD2fficPOREQeymi23hM0NGNBsCvXOHXqBLp3fxAtW7a0lPXv\nH2r5d2HhEeTmZmH//o8sZbW1tSgq+hzx8Y8gK+sg0tPTcOLEcdy4cQPV1b+htPQf6N9/AE6dOoGl\nS5+y/F5ISP8m6zJz5nR88UURAODmzRsAgPfeewcAMGzYcPz97x/av6FmYPAlIvJQOo0ZBivl+maM\nibpyjbCwgfjuu3/h+vXr8PGpS/N55kwx+vbtBwAYNSoCbdq0xapVaxr9bnV1NZ56agneeGMd1q9/\nB/fc44vBg0MsvduwsEEoKTmDiRMnW67blLS03ZZ/r1r1KjQaDZ5++ln7N+EkDjsTEXmopMRJuFqa\n0aDsaukezJg6UZRrdOrkj969+2DlyldQUVGBd97ZgPLyny2PT5mSgIMH9+PgwUxUV1ejuroaubnZ\nqK7+7dZ/1fD394fJZMLatW/iypUyy+/GxMRh7949+OKLYygt/Qd27Njm8D0B7l9xzZ4vEZGHql+R\n/OHeAzCY6nqrqfNnNGu1s6vX2LhxK9566w1ERoZj3LiJmDJlquWxNm3a4sMP92Hnzu149tmnYDab\nMWzYCIwYMRKdOvnj+edfwhNPLMDNmzeRnDwL4eHDLb8bExOHqqoqLFu25NZq5yU4ceJLh+/L3Yuz\neKqRDKj9VBUhsa0cw3ZynNrb6ttvzwMAD1YQmK12dfRUIw47ExERiYzBl4iISGQMvkRERCJj8CUi\nIhIZgy8REZHIGHyJiIhExuBLREQkMgZfIiIikTHDFRHZdTT/EHLSt0NrMsGk1SIuaRZGRo+Rulqk\nQkuWLELnzl3w7LN/FvS6Fy9+jyFD+qOs7BdotdL3Oxl8iahJR/MPIXP9asxp3xGABjCZsXn9agBg\nACbBaTQaWZ276y7Sh38ikrWc9O23Au9tc9p3RM6uNIlqRGonp3N33YU9XyJqktZkAtC4J6I1GMWv\nDAmu+EwJ8gqLYDCaoddpEBMRjtD+IaJd49Kli3jrrTeQlXUQY8eOh9HY8IDC0tJ/YMuWjcjLy8Hk\nyY/g8cf/iC5dugIA1q1bje3bt6Gq6hdER8ciJWUuhg0bAaAugO/duxtr1qy6dbDC4mbdk7ux50uK\ncTT/EJ6eNRMrkpLw4oI5OJp/SOoqeQSTjfkxk14nck1IaMVnSrBz/xHoA4agZeeh0AcMwc79R1B8\npkS0a8ydOxu+vr4oLCxCjx498fHH+yzDzlevXkVCwkRER8eisPAY2rVrj4UL51p+t3v3B5GZmY3T\np88hNDQMixbNszyWm5uN1177G958cx02bHgXmze/L6vhbAZfUoT6eccZNw2YbjRjpsmMzPWrGYBF\nEJc0C5sryxuUbaooR9yjMyWqEQklr7AIgUERDcoCgyKQV1gkyjV+/vlnfP31V3j22RfQvn17PPFE\nKjp27GR5/MCB/YiPn4Lx4yfC1/deLF68FBcu/As//1x35m98/CPo1MkfPj4+WLjwT9BoNDhz5nRd\nvfJyMG3aowgPH47g4H6YPXuOrIazGXxJETjvKJ2R0WMwackypOm12Akz0vRaxKcu42IrFTAYrQcj\nW+VCX+PUqRPo3v1BtGzZ0lLWv3+o5d+FhUeQkbELPXt2Q8+e3RAU1B3V1b+jqOhzAEBW1kH84Q8z\nERLSC7163Y+ffrqC0tJ/WK7dr9/toe+QkP4267Fnz4fo3v0+dO9+H/7jP6bZrbcQOOdLisB5R2mN\njB7DYKtCep31YVhb5UJfIyxsIL777l+4fv06fHzqzlU+c6YYffv2AwCMGhWBNm3aYtWqNY1+t7q6\nGk89tQRvvLEO69e/g3vu8cXgwSGW3m1Y2CCUlJzBxImTLde1Zdq0GZg2bYbd+gqJPV9SBM47Egkv\nJiIcZecKG5SVnStATES4KNfo1MkfvXv3wcqVr6CiogLvvLMB5eU/Wx6fMiUBBw/ux8GDmaiurkZ1\ndTVyc7NRXf3brf+q4e/vD5PJhLVr38SVK2W36xUTh7179+CLL46htPQf2LFjm8P3JAYGX1IEzjsS\nCS+0fwiSJ4+G4cpx3PjhCxiuHEfy5KhmrXZ29RobN27Fr7/+gsjIcJw//09MmTLV8libNm3x4Yf7\ncPRoIYYPH4hhw8Kwa9dOAHWB+/nnX8ITTyxAVNRI1NbWIjx8uOV3Y2LisHz5c1i2bAkWL16IlJT5\nslpwpTG7eQa6psaAqqrr7nwJxfPzqxtuYTs17Wj+IRzOSIfGYECNRoO4R2dyKNQGvqccp/a2+vbb\n8wCAHj0ekrgm6mKrXf38fODtbX9Gl3O+pBgjo8dgQsIkAOr9oCQiz8BhZyIiIpEx+BIREYmMwZeI\niEhkDL5EREQiY/AlIiISGYMvERGRyFwKvpcuXUJUVBSCg4MxevRo7NixQ6h6ERERqZZLwdfLywtr\n1qxBaWkp9uzZgz//+c+4du2aUHUjIiIPlp6ehvj4sZafu3e/Dxcvfi9hjYTjUvANCAjAgAEDAAAd\nOnRAcHAwvvzyS0EqRkREdKfvvvsR3brdL3U1BCFYhqvz58+jtLQUQ4cObfgCep0lfRtZp791OADb\nyT62lWPc2U5HcnKQuW0rtAYjTHodJj2WgtFxcYK/jljU/p7y9W2Ba9duSl0NxRg0qB8+/vhTdOnS\n1e5zfX1bNHrf6B087EWQ4Hvt2jXMmDEDa9asQevWrYW4JBHJ0JGcHOxZ+TpS2naoKzAasHXl6wCg\n6ADsyY7mH0JO+nZoTSaYtFrEJc1qds50V66xbt1qbN++DVVVvyA6OhYpKXMxbNgIq8/19/dDUdFp\nVFZWIiUlGSUl/7QclnDgwCdYtepVHDlSd9bvoUM52LLlA5w//w1SUuZh9uwUtG59j936iHX4gsvB\nt7a2FomJiZg9ezamTJnS6HGDwcg8vHaoPbG7kNhWjnFXO330wabbgfeWlLYdkLZ5M8LCHxb0tcSi\n9vdUU73eo/mHkLl+Nea07whAA5jM2Lx+NQA4HDxdvUb37g8iMzMbvr73YuvWTVi0aB5On/6qyd8Z\nNGgwWrVqjcLCI4iMjAIA7N27G9OmPQoAyMo6iNde+yveeGMtunbtij//+VlcvVqJ//qv/+PQPTnq\n2rWbjd43jh6s4NKcr9lsxrx589CvXz8sXbrUlUsRkQJoTSbr5QajyDUhIeSkb78VNG+b074jcnal\niXaN+PhH0KmTP3x8fLBw4Z+g0WhQXHzK7u9NnToN+/btAQD89ts1HDqUi4SEaQCAjz/eiyVLlmLw\n4CHw9w/Ak08uw6efZjp8T24+7A+Ai8H36NGj2L59O/Lz8xEWFoawsDBkZWUJVTcikhmT1vpHhsnB\neS6SFyG+TLl6jaysg/jDH2YiJKQXevW6Hz/9dAVnz5ba/b2EhGk4cOAT1NTU4MCBTxAaOgCdO3cB\nABQWHsaKFcvQs2c39OzZDQkJk3Dp0kWUl5c3us7ly5csz+vZsxsuX76E0aNHWH6uD/BCc2nYedSo\nUTDZaHgiUp+4pFnYbBlirLOpohzxqcskrBU5y6TVAqbGvbzmfJly5RrV1dV46qkleOONdVi//h3c\nc48vBg8Ocajn2bt3ELp06YpDh3KRkbEbU6dOtzw2alQkJkyYhClTptq9TpcuXfHNNxctPw8eHIKP\nPjro0IIrVzDDFRE5bGT0GExasgxpei12wow0vRbxqcuavUCH5CEuaRY2VzbsDW6qKEfcozNFuUZ1\n9W+orq6Gv78/TCYT1q59E1eulDn82omJ0/Hee2+jqOhzTJ78iKV8+vQkbNiwFkVF/wuj0YiKigpk\nZR10+LpiEGyrERF5hpHRYxhsVaL+/2ParjTL1rHmfply5RqdOvnj+edfwhNPLMDNmzeRnDwL4eHD\nLY9rNJoGq4/vXomckDANf/3rS4iJiUPbtu0s5WPGxAIAPvjgPcyenYQ2bdoiISER48ZNcPi+3E1j\ndvPMck2NQfGrCPMLCpCekQmjWQOdxoykxEmIjowU7PpqX20pJLaVY9hOjlN7W3377XkAQI8eD0lc\nE3Wx1a6OrnZmz9eO/IICrNu4C+2CEwEABgDrNu4CAEEDMBEReQ7O+dqRnpFpCbz12gUn4sO9BySq\nERERKR2Drx1Gs/VsJwYu8iYiIicx+Nqh01ifEtez5YiIyEkMIXYkJU7C1dKMBmVXS/dgxtSJEtWI\niIiUjguu7KhfVPXh3gMwmOp6vKnzZ3CxFREROY3B1wHRkZEMtkREJBgOOxMREYmMwZeIiGQpPT0N\n8fFjLT93734fLl78XsIaCYfDzkREpAjfffej1FUQDHu+REREImPPl4jIg50tPoWTebmAsRbQeWFg\nTCz6hoaJdo1161Zj+/ZtqKr6BdHRsUhJmYthw0ZYfa6/vx+Kik6jsrISKSnJKCn5p+WwhQMHPsGq\nVa/iyJHPAQCHDuVgy5YPcP78N0hJmYfZs1PQuvU9zbovd2LPl4jIQ50tPoVj6TsQ5eWNqJatEeXl\njWPpO3C2+JRo1+je/UFkZmbj9OlzCA0Nw6JF8+z+zqBBg9GqVWsUFh6xlO3duxvTpj0KAMjKOoj/\n+39fwpNPPo2PPjqIEyeOY+3a1Q7fkxgYfEm1zhafwvY3V2L7yr9h+5srm/WBQuQJTublYmxg5wZl\nYwM71/ViRbpGfPwj6NTJHz4+Pli48E/QaDQoduBvderUadi3bw8A4LffruHQoVwkJEwDAHz88V4s\nWbIUgwcPgb9/AJ58chk+/TTT4XsSA4MvqZIQ3+iJVM9Y27xyN1wjK+sg/vCHmQgJ6YVeve7HTz9d\nwdmzpXZ/LyFhGg4c+AQ1NTU4cOAThIYOQOfOXQAAhYWHsWLFMvTs2Q09e3ZDQsIkXLp0EeXl5Y2u\nk5SUiO7d70P37vdh797dDtVZCJzzJVWy9W38cF5us+ezxFB8pgR5hUUwGM3Q6zSIiQhHaP8QqatF\naqfzal65wNeorq7GU08twRtvrMP69e/gnnt8MXhwCBw5Zr537yB06dIVhw7lIiNjN6ZOnW55bNSo\nSEyYMAlTpky1e5309Ay7z3EH9nxJnYT4Ri+S4jMl2Ln/CPQBQ9Cy81DoA4Zg5/4jKD5TInXVHFJ8\npgRvbtiI19e+jzc3bFRMvQkYGBOL7LIfGpRl/3gZA2NiRblGdfVvqK6uhr+/P0wmE9aufRNXrpQ5\n/NqJidPx3ntvo6joc0ye/IilfPr0JGzYsBZFRf8Lo9GIiooKZGUddPi6YmDPl9RJiG/0IskrLEJg\nUESDssCgCOQVFsm+91v/xSEwKMLyYbJz/xEAkH3dCZZRoMN3rFQeljyzWaNDrlyjUyd/PP/8S3ji\niQW4efMmkpNnITx8uOVxjUZjWc1c//OdEhKm4a9/fQkxMXFo27adpXzMmLrA/8EH72H27CS0adMW\nCQmJGDdugsP35W4asyP9exfU1BhQVXXdnS+heH5+PgDAdnKAo21VP+d759Bz9o+Xm/3BIobX176P\nlp2HNiq/8cMXeObJx526pljvqTc3bIQ+YEijcsOV43hq8Xy3vrZQ1P739+235wEAPXo8JHFN1MVW\nu/r5+cDb236/lj1fUiUhvtGLRa/TNKtcTgxGs9UPEYPRrd/piRSPwZdUq29omCyD7d1iIsItQ7f1\nys4VIHlylIS1coySvzgQSYnBl0hi9XOjd652Tp4cpYg5UyV/cSCSEoOvBxMirRwJI7R/iCKC7d2U\n/MXBk1y8eEHqKqjOxYsX0K3bA07/PoOvh2qwIMnLGwCQnb4DABiAqVmU+sXBUzzwQHdBruPr2wIA\ncO3aTUGup3Tduj3gUtsy+HoopSWhICLn6HQ6QVY6q31VuNiYZMNTKSgJBRGR2jD4eioFJaEgIlIb\nBl8PJURaOSIicg7nfBUuv6AA6RmZMJo10GnMSEqchOjISLu/p6QkFEREasPgq2D5BQVYt3EX2gUn\nAgAMANZt3AUADgdgBlsiIvFx2FnB0jMyLYG3XrvgRHy494BENSIiIkcovufr7LCrGhjN1lP4GUwi\nV4SIiJpF0cHX1WFXpdNpzDBYKddzPIOISNYU/THt6cOuSYmTcLU0o0HZ1dI9mDF1okQ1IiIiRyi6\n5+vpw671vfsP9x6AwVTX402dP8Mjev1EREqm6ODLYde6AMxgS0SkLIoOUxx2JSIiJVJ0z5fDrkRE\npESKDr4Ah12JiEh5XBp2njt3Lvz9/RESwrM8iYiIHOVS8J0zZw6ysrKEqgsREZFHcCn4Pvzww2jb\ntq1QdSEiIvIIbp/z1et18PPzcffLKJperwMAtpMD2FbAiVPFyDr0OWqNZnjpNBg3ZgQGhYU2eA7b\nyXFsK8ewnRxT3052n+fmehCRgE6cKsbW3bkI7B2JFrfKtu7OBYBGAZiI5MvtwddgMKKq6rq7X0bR\n6r9Jsp3s8/S2+vhgITr0GIVag9FS1qHHKHx8sBAPPdjLUubp7dQcbCvHsJ0c4+fnA29v+6FV0Uk2\niDyNwWhuVjkRyZNLPd/k5GQUFBSgsrISXbt2xcsvv4w5c+YIVTeiZvGE4yX1Ouv5zG2VE5E8uRR8\nd+7cKVQ9iFziKcdLxkSEY+f+IwgMirCUlZ0rQPLkKAlrRUTNxQVXpApNHS+ppuAb2r8uoU1eYREM\nRjP0Og2SJ0dZyolIGRh8SRU86XjJ0P4hqgu2R/MPISd9O7QmE0xaLeKSZmFk9Bipq0XkNgy+pAo8\nXlK5juYfQub61ZjTviMADWAyY/P61QDAAEyqxY8mUgV3Hi+ZX1CABanLMW/JCixIXY78ggKXr0m3\n5aRvvxV4b5vTviNydqVJVCMi92PPl1TBXcdLespCLilpTSYAjacNtHfsZSZSGwZfUg13HC/pKQu5\npGTSagFT433KJgfT9BEpEYediZrgSQu5pBKXNAubK8sblG2qKEfcozMlqhGR+7HnS4p0tvgUTubl\nAsZaQOeFgTGx6BsaJvjrcCGX+9UvqkrblQatwQiTXof41GVcbEWqxuCrUmrO9nTmxEkcS9+BsYGd\nAS9vAEB2+g4AEDwAJyVOajDnC9Qt5EqdP0PQ17lT8ZmSBvt4YyLCVbe16G4jo8cw2JJHYfBVIbUv\nEvoi69O6wHuHsYGdcTgvV/Dg666FXLYUnymxZLCq/+Pcuf8IAKg+ABN5EgZfFVL9IqHaWkDXonG5\nsdYtL+eOhVy25BUWNUgdCQCBQRHIKyxi8CVSEQZfFVL9IiEvL8Davei8BLm8FEP29RmeLl+8DGOL\nLHQaPBad+wy1PK7UU4uYuYrIOgZfFVL7IqGh48Yj+4NNDYaes3+8jGHJrq+OrR+yN/sFofz709Bq\ndXjmpVWYmViCpYsXu3x9axpkeOpSd09rCnfhB8ASgJV4apFcM1fd/YXgkXlzMTouTrL6kGdSyccx\n3cmd2Z7koP+ggRiW9B84XFuDwzeqcbi2BsOSZwoy35uekVkXeC+cQtDImeg1PAkhY5dhV+b/uC2z\nlbUMT/95f2f8fCIHQN2pRTER4W55bXeSY+aq+i8EM01mJEODmSYz9qx8HUdyciSrE3km9nxlICfv\nMLbu/Ag3a02CDHOKvUhICn1Dw9yytcho1qD8+9MIGjWrQXnQ6IVumzO3leFJd+NXGK4cV+ypRXLM\nXGXtC0FK2w7YtW0bwsIflqhW5IkYfCWWX1CADZt2o02fqQCEW5ks5iIhNdFpzNBqrWdWcsec+dni\nU/jp8iXkQgOT2Yyefn548F4/AECXB7riqcXzhX9Rkcgxc5WtLwQag7WJGiL3YfCVWHpGpiXw1lPV\nymSBnTlxEl9kfYqbv193S3KNpMRJeOalVVYfE3rO/GzxKRxL34E/hg/HT9/9C928WyDr0kUAwJGa\nGsSnLhP2BUUWlzQLmy1zvnU2VZRLel+2vhCY9cr4KOQCNvXgnK/EVL8yWUBni0+hcNtWROj0iGrZ\nGlFe3jiWvgNni08J9hrRkZGYmTgO546826DcHXPmJ/NyMTawM9q17wj/7g/isgYI6doV+6/9WxUZ\nnkZGj8GkJcuQptdiJ8xI02slvy9rqSy3/FKBiY89JlGNHGdtvjpz/WoczT8kddXICcr4uqdial+Z\nLKSTebmIC7yvQZkjyTWau3Vo6eLF6B8S4v45c2OtJUNXu/Yd0e5WD3FEr16KD7z17unQCff2G2HJ\n1nVPh06S1sdaKsvpK57B6Lg4VFVdl7Ru9thawJa2K0017xdPwuArsaTESQ3mfAH3py9ULGMtgJY2\nyq1zNtuXKHPmtvYlC7Rf+cSpYnx8sFCyNJVyzdZ1dypLPz8fyerSHHJcwEbOY/9KYtGRkXhm8Sz4\nlOdA80MWvMqyVLcyWTB3BKWKykqcKT2L4n+cRW7h5za3ATWV7UtqA2NikV32Q4Oy7B8vY2BMrMvX\nPnGqGFt350IfMAQtOw+FPmAIdu4/guIzJS5f21FNZeui5jNprX9c8+hFZWLPVwbiYqIQFxMl+2Ev\nqQ2MiUVOxocI07fAdxd/hJdvIA5evoSaXo/Y7M3KeU69fqj88B2nMwm1Xznr0OcI7B2J2jt6RWKn\nqTQYzVY/YJSarUtqclzARs5j8CXF6BsahtatW+LF/1wObatuqP39VxgeGAWfTj3g06mH1RXicp9T\nd9d+5VqjGVayX4sa+Gxl5VJiti454NGL6sLgS4rSf9BAtOjRD7UBcfACcOfsqLXerBRHAsqBl53A\nJ8aWlZiIcMucb72ycwVInhwl6Ot4Eh69qB4MvqQ4Oi1gbYmVtd6sJ2T7smbcmBHYujsXHXqMspTV\nBz6xci7XD2/feTaxUrN1EQlNYzab3ToOVVNj4FymHfWrLdlO9vn5+SAn7zBeXfd3q71ZtQdVR/n5\n+dhc7fzigjmYaSXRRJpei7/89yYJaist/v05hu3kGD8/H3h72+/XsudLihMXE4Xq3296XG+2uQaF\nheKhB3s1Km/ulhVmVSISHoMvKZKac1e76zzh+iB68etzeM9kwtBO/hjQvoPlcWtbVuR6LCDZxi9L\nysDgSyQjziYFsadBEO3dBwDw3lelAIAB7TvY3LLCrErKwi9LyiGTDRdEBLgvKYi1ILqgTzD2l//U\nZM7luiHqxphVSZ7keIYyWceeL5GMuCspiK153p5BffFiE4us5HgsINnGFJS2yW04nj1fIhnRaaxv\nPnA1KYizqQmtnQK0qaIccY/OdK1C5BZMQWmdHE+EYvAlkpGkxEm4WprRoEyI4wydDaJyPBaQbOOX\nJevkOBzPYWciGXFXUhBXUhMyq5JyMAWldXIcjmfwJZIZd22jYhD1DPz/3Jgc1y5w2JmIiFRNjsPx\n7PkSEZGqyXE4nsGXiIhUT27D8Qy+JDvuSq9IRCQXDL4kK+5Kr0hEJCdccEWy4q70ikREcsKeL8mK\nu9IrkrzJLfWfFIrPlCCvsKjR+cukTi4F38LCQixcuBAGgwGpqalYsmSJUPUiD6XTmGG49e/Ky6Uo\n//40tFqHHUYiAAAaEElEQVQdzL9fQX5BARImj5O0fkpwtvgUSv/nMFBbi5smDQbGxKJvaJjU1bKJ\nJ/HUBd6d+48gMCjC8qG8c/8RAGAAVimXgu+TTz6Jd999F/fffz/Gjh2L5ORkdOjQwf4vEtmQlDgJ\n6zbugtkvCOUXTiFo1CzLY+s27kLrVi0QFxMlYQ3l7WzxKRxL34GJ3boCOj0MtSZkp+8AANkGYB5b\nCOQVFiEwKKJBWWBQBPIKi1QVfDnCcZvTwbeqqgoAEBFR94aJi4tDUVERJk50LQctebb6RVUv/PVN\nBI1Z2uCxdsGJ2Jb+kcPB1xNXTZ/My8XYwM4NysYGdsbhvFzZBl85pv4Tm8FotvphbDBaP2hDiTjC\n0ZDTwff48eMICgqy/Ny3b18cO3asUfDV63Xw8/NxvoYeQH8rxRnbqU7C5HH4cN+nqLXymNGkceg9\nlZN3GBs27UabPlMB1K2aXrlhG95+fwvad/CHTgukJD+iul50C60Zei8ttJq6YKb3qltT2cJolu37\nS9vCC7hpaFSua+EtSp3l8PfXupUXtFZSHepaecnm/5ur7ZSfsdPqCMeujHRMSJjkcv3kQu9gykou\nuCJZ0mlhNfg6mop1686PYPTtjXNH06DV6mAyGdHx/gEo+/407u0Xh1oAr2/YDgCSBOAzJ07ii6xP\ngdpawMsLQ8eNR/9BA12/sJdX88plYNJjKdi68nWktL09ZbXllwpMX/GMhLUS17gxI7B1dy4Ce98e\nmSn7+ghSpsdJWCth2RrJ0Bgaf/HyBE4H3yFDhmD58uWWn0tLSzFuXOPFMAaDEVVV1519GY9Q/02S\n7XTbtCnjG+z3BeqO1luY+phD76kffixD+W8/N5gzPvfZdtRc/7fl5zZ9pmLT9r0IHzJM+BtoQv28\n7NjAzoCuBWACsj/YhOrqGy4PDQc/HIUD9XO+QN2c74+XMSx5pmzfX2HhD+P3J2oapP6b+MR/Iiz8\nYVHqLIe/v4ce7IVp428ir/CYZbXztPGReOjBXrL5/+ZqO9XYGEGv0Whkc49C8PPzgbe3/dDqdPD1\n8/MDULfiuVu3bsjNzcWLL77o7OWIGrB1tJ6jvdSKyl8azRkHjZqFE5lvNCiTYguTO+dl63+/8I7V\nzsOSZ8p2vree3FL/SSG0f4iqFlfdLS5pFjZb5nzrbKooR3zqMglrJR2Xhp3feustLFy4ELW1tUhN\nTeVKZxKUK0frBQQEWC1veU/bBj/rpUgzY6wFvLytlwugb2gYhkeMAMDRFJIPOR5uICWXgm9kZCS+\n+uoroepCJJj2bf1gbSZJ79XS8u+rpXuQOn+GeJWqp7Mx/2qrnEglOMJxG9NLkiolJU7C1dKMBmWX\nj/8d/r4GaH7IgldZFlLnz5Bk69HAmFhkl/3QoCz7x8sYGBMrel2ISBpc7UyqZG3O+NnUP9gNtmLs\nDa6ffz2cl1s31KzzUsS8LBEJR2M2m926i7umxsB5JzvksNpSKdzZVnefqAQAV0szkDr/UacD8Nni\nUzh5R5AVK9Uj31OOY1s5hu3kGEdXO3PYmegWoU9Uqt9SFOXljaiWrRHl5Y1j6TtwtviUENUlIgVj\n8CW6RegTlWxtKTqZl+vcBYlINRh8iW7RaazPwDi9HcnW1iGBthQRkXJxwRXRLfUnKt2dVcvp7UgS\nbynKyTuMD/6e4VEHSxApBYMv0S22smo5G7AGxsQiuz6N5C31qR7dLSfvMF7fsL3BwRLrNu4CgCbv\nR6oFYu7CI+xIrrjaWQa4itBxcm4ra9uUAtrcK0kw+9OyZ3GjU+Ok/F5lWXh37Uqrdd/0wVa0qSjD\nxK5dcV+APzq0b4/ssh8wLOk/FBmAGx5hV2dzZTkmLWmYVUnO7yk5YTs5xu25nYnotru3KdX3NFPn\nP4pZT60QvT5GG4vErC0eq6+7b00rjH9oEEwAvrv4IwD5nwXclJz07VaPsEvblcbeL0mOC66IBCD0\nNiVX6Wz8ZVtbPFZfdy/cjsxevoH48cpPdT8odIGY1mT9G4ito+2IxMTgSyQAobcpuSol+RH8+tXe\nBmVXS/dgxtSJAOp6uwtSl2PekhX46p//AgDU3vVxYJmQUmjOaZPW+sebydFDoYnciMPOErh7bnDe\n7ERJDnQn4eg0ZusHOUj09bb+/bRp+95Gi8fuHiI3XkgDABg6h+Lg90cxoXMXAIBGI94CMXfgEXYk\nZwy+IrM2N/j6hu0AIPqh7iQcwbcpCSAuJsrqe+ruIfKO9w/Auc+2I2jULFwFkP5DMYxXv0G/fr3x\niIJzTvMIO5IzBl+RWZsbbNNnKralf8Tgq2BCb1Nyp7uHyNt3CQYAfH1oLYKCekMf0Aoz/vRfsqx7\ncwlxhB23K5E7MPiKzObcINeAWKWkIfroyEjBA5Y7TlmyNkTevkswAnSXrG5D8mQNtytpAJMZm9ev\nBgAGYHIJg6/IbM4Ncg1II54+RG9r+xLQMFGGtcQYwyNG2LyuHIfI5YrblchdGHxFZu2D79ezGVi4\nZLaEtZInTx+ib2r7Un3wrT85aWxgZ8DLGwCQnb4DrVu3RP9BA61eV0lD5O5y51CytoUXJj2WgrDw\nhxs9r267UuPRKim3K3EYXB0YfEVm7YPvmSWzERcTxcwxd7E3RC/GwfdScmT7kq2Tkz7P+tRm8AXc\nM0SuFI2Gkm8asHXl6/j9iZpGQcyk1QKmxkkApdquxGFw9WDwlcDdH3z1aduooaaG6B0dklUyh7Yv\nGWstPd4GapWZGEMM1oaSU9p2sDqULLftShwGVw8m2SDZSkqchKulGQ3Kfj2bgceSHpFdRil3sHb/\ndybKAGA7AYaXMhNjCKn4TAne3LARr699H29u2IjiMyUAmpf5amT0GExasgxpei12wow0vVbS7UrM\n2qUe7PmSbDU1RP/etr1Wf0eqjFLu4MjcrK2Tk8bMnyd6feWk+EwJdu4/gsCgCMuH3M79RwA0fyhZ\niO1KQpHbMDg5j8GX3EaIOVlbQ/RyyyjlLvbmZusTYBy+Y7XzsOSZTc73eoK8wiIEBkU0KAsMikBe\nYZHVoeQtv1Rg4hP/KXY1m01uw+DkPAZfcgt3z8lyu8xtfUPDFJuFyl0MRrPVDzeD0dwo85WuhTem\nr3jG6mpnuWHWLvVg8CW3cGSbjCu4XYaaotdZXyleX37nULLSzqmV0zA4OY/Bl9xCjFN+PHm7DDUt\nJiLcMudbr+xcAZInC5MdjXttyVUMvuQya3O7njInS/IU2j8EQN3cr8Fohl6nQfLkKEu5K7jXloTA\n4EsusTW3Ozq8D44UZXBOliQT2j9EkGB7N+61JSEw+JJLbM3tfvVNFlLnP8o5WVIdOaacJOVh8CWX\nNDW3a2tO1tpBAEpfravGeyLruNeWhMDgS02yt1e3uXO7tg4CAKDYYOXIPak9D7Un4V5bEgKXv5BN\n9fO5hvvGw9x5HAz3jce6jbuQX1BgeY5DKRDvYOsggJN5ucLfgEjs3ZMj7UjKIbeUk6RM7Pm6QO29\nGXv5k+vvXW++gV/PbEXb9v7253ZtHQRgVPBBAHbuyd17nkl83GtLrmLwdZInnKpjaz73yk8/Nbj3\nezsDV0szMHPaRPv3busgAFvlSmDnnsTY80xEysJhZyd5wqk6Ok3jRSUAUFH5i9P3PjAmFtllPzQo\ny/7xMgbGxDpfUYnZuydb7ajmPc/5BQVYkLoc85aswILU5RxiJ7oLe75O8oTejK38yQEBAVaf78i9\n2zoIQKmLrQD792SrHUeH98WC1OWqm7bwhFEhIlcx+DrJEzI42cqfnJ6R6dK9q/EggKbuyVo7jg7v\niyNFX6kyQHGOm8g+Bl8necqpOrb26nrCvQvp7nZckLpctQHKE0aFiFzF4OskNZ2q09xV22q69+YS\naoW7mgOUJ4wKEbmKwdcFajhVx9n5OTXce3MJOZep5gClllEhnlxE7qSCP3VyhSes2haKkG3V3OQk\nShIdGYnU+Y/CqywLmh+y4FWWpbiRkfqTi2aazEiGBjNNZmSuX42j+YekrhqpBHu+Hk7Nw59CE7Kt\n1D50r/SREZ5cRO7mdPDdvXs3XnrpJZw7dw7Hjx/HwIEDhawXiUTNw59CE7qtXA1Qas+wJiWeXETu\n5vRHbEhICPbt24eIiAgh60MiU/Pwp9Dk1FbMF+1eJq31j0aeXERCcbrnGxQUJGQ9SCJqH/4Ukpza\nintp3YsnF5G7uX3OV6/Xwc/Px90vo2j6W9+mpWqnhMnjkDB5nCSv3VxsqzoarY0emEYLPz8fydtJ\nSay11YSESWjV2hu7tm2DxmCAWa/HjOeexei4OKmqKTm+pxyjd3B0pMngGxsbiytXrjQqf+WVVxAf\nH+9czYjIZTotYO0cKI6KCmd0XJxHBdsjOTnI3LYVWoMRJr0Okx5L8aj7F1uTwTc31/UzVg0GI6qq\nrrt8HTWr/ybpaDt58kKb5raVWvV6sAt27v9vBEcvspSV5r+D5MmRqKq6znZqBrbV7a1VlmF2owGb\nX3sNv1fXWFZ3s50c4+fnA29v+4PKggw7m83WT20h4TFpPQHA2X9+j4BeD+Pro2nQaHUwm4wI6BWB\nr765KHXVSIG4tUp8Tgffffv2ITU1FRUVFZg4cSLCwsLw6aefClk3soILbZTB3aMTRrMG7bsEo32X\n4Ablhh8uCfYa5Dm4tUp8TgffhIQEJCQkCFkXcgCTYsifGKMT3J9NQjJptYCp8Qgmt1a5D/9UFcYT\nD2ZXGjFSdsppzzEpX1zSLGyuLG9QtqmiHHGPzpSoRurH9JIKo5ak9Urk6FCyGKMTctpzTMpXP6+b\ntivNsto5PnUZ53vdiMFXYfihK43mDCWLNSSs9PzJJC8jo8cw2IqIwVeB+KErvuYsdOPoBBHZw+BL\n5IDmDCVzdIKo+Tzt/GQGXyIHNHcomaMTRI5rmORDA5jM2Lx+NQCoNgAz+BI5QIqhZE/OZOYu1npX\nExImSV0tj+eJST4YfIkcIPZQMjOZCc9W76pVa2/mMJaYJyb5YPCVIfZ4hCVUe4o5lMxMZsKz1bva\ntW0bg6/EPDHJB1MzyAwPSReWUtuTmcyEV9e7akxjsDabT2LyxCQf7PnKDHs8wlJqe0qRPlLtIy62\neldmPT8GpeaJST74rpMZWz2es19/i/yCAlV9GIpBqT1IsRd4ecIcc1zSLGy+89g81PWuZjz3rFPX\n87StMe7maUk+GHxlxlaPx+TVRnUfhmKo+qUS93ZuXC73XNhiL/BS6ghBc9jqXTkz3+uJW2NIWAy+\nMmOtx3Pus7+j4wMD0a5LsKo+DN0tv6AAV6uq8eNn2xE0apal/PLxv+PZ1D9IVzEHibnAS6kjBM0l\nVO/KE7fGkLAYfGWm/sP2hb+uhqaVP8wmIzo+MNBybqvaPgzdKT0jEw+MXIDKy6UNDp339zXxC8xd\neERh83ji1hgSFoOvDEVHRiI9IxOG+8Y3ekyNH4buWuhT35u7+9B5zQ9ZLl9bbZiPunk8cWsMCUuF\nH+Xq4CnntbpzKxDPPnZcdGQkUuc/Cq+yLGh+yIJXWRbzUTfBE7fGkLDY85UpT0nO786FPuzNNQ/z\nUTvOE7fGkLAYfGXMEz4M3bnQx1O+wJA0PG1rDAmLwZck5e6FPp7wBYaIlIfBlyTFoWFSiuIzJcgr\nLILBaIZep0FMRDhC+4dIXS1SKAZfkhSHhkkJis+UYOf+IwgMirB8aO7cfwQAGIDJKQy+JDkODZPc\n5RUWITAookFZYFAE8gqLGHzJKdx0QURkh8FofduarXIiexh8iYjs0Ousr8q3VU5kD4MvEZEdMRHh\nKDtX2KCs7FwBYiLCJaoRKR3nfImI7Kif171ztXPy5CjO95LTGHyJiBwQ2j+EwZYEw2FnIiIikTH4\nEhERiYzBl4iISGQMvkRERCJj8CUiIhIZVzuT6PILCpCekQmjWQOdxoykxElML0lEHoXBl0SVX1DQ\n4BQjA4B1G3cBAAMwEXkMDjuTqNIzMhscHwgA7YIT8eHeAxLViIhIfAy+JCqj2XouXINJ5IoQEUmI\nwZdEpdNYPwVGz3ciEXkQfuSRqJISJ+FqaUaDsqulezBj6kSJakREJD4uuCJR1S+q+nDvARhMdT3e\n1PkzuNiKiDwKgy+JLjoyksFWZNzeRSQvDL5EKsftXUTy4/Sc7/Lly9GnTx8MHDgQS5cuxfXr14Ws\nFxEJhNu7iOTH6eAbFxeH0tJSfPnll6iursaOHTuErBcpVH5BARakLse8JSuwIHU58gsKpK6Sx+P2\nLiL5cTr4xsbGQqvVQqvVYuzYsSjgh6zHqx/eNNw3HubO42C4bzzWbdzFACwxbu8ikh9B/vzef/99\nxMfHC3EpUjAOb8oTt3cp29H8Q3hxwRz8ZX4KXlwwB0fzD0ldJRJAkwuuYmNjceXKlUblr7zyiiXY\nvvzyy/D19cX06dOtv4BeBz8/HwGqql56vQ4AFN9OGq3O1gOC3Zta2srd7mynhMnj0LpVC2xL/wgG\nI6DXAc+lPoa4mCiJaykPcn5PHcnJwcH/twYpbTsA0AAmM7b+vzVo1dobo+PiRK2LnNtJTurbye7z\nmnowNze3yV/esmULsrOzcegQv4kRoNMCtVbKHXwvkhvFxUQx2CpQ5rattwLvbSltO2DXtm2iB18S\nltNbjbKysrBq1SoUFhaiZcuWNp9nMBhRVcWV0E2p/yap9HaaNmV8gy0tQN3wZur8GYLdm1rayt3Y\nTo6Tc1uZbtYCaLxgznizRvT6yrmd5MTPzwfe3vZDq9PBd8mSJaipqUFMTAwAYPjw4Xj77bedvRyp\nALNXEQnLpNUCpsYL5kwcTlI8p4PvN998I2Q9SCWYvYpIOHFJs7B5/WrMad/RUrapohzxqcskrBUJ\ngRmuiIhkamT0GABA2q40aA1GmPQ6xKcus5STcjH4EhHJ2MjoMQy2KsRt9kRERCJj8CUiIhIZgy8R\nEZHIGHyJiIhExuBLREQkMgZfIiIikTH4EhERiYzBl4iISGQMvkRERCJj8CUiIhIZgy8REZHIGHyJ\niIhExuBLREQkMgZfIiIikTH4EhERiYzBl4iISGQMvkRERCJj8CUiIhIZgy8REZHIGHyJiIhExuBL\nREQkMgZfIiIikTH4EhERiYzBl4iISGQMvkRERCJj8CUiIhIZgy8REZHIGHyJiIhExuBLREQkMgZf\nIiIikTH4EhERiYzBl4iISGQMvkRERCJj8CUiIhIZgy8REZHIGHyJiIhExuBLREQkMgZfIiIikTH4\nEhERiYzBl4iISGQMvkRERCJzOvi+8MILCA0NxYABAzB79mxUVlYKWS8iIiLVcjr4rlixAsXFxTh9\n+jR69uyJtWvXClkvIiIi1XI6+Pr6+gIADAYDqqur0bJlS8EqRUREpGYas9lsdvaXn3/+ebz77rvo\n3bs3Dh8+DG9v70bPMZnMMBiMLlVS7fR6HQCwnRzAtnIM28lxbCvHsJ0co9froNVq7D6vyeAbGxuL\nK1euNCp/5ZVXEB8fDwD4/fff8fzzzwMA1qxZ42x9iYiIPIZLPd96JSUlePzxx3Hs2DEh6kRERKRq\nTs/5fvPNNwDq5nx37tyJqVOnClYpIiIiNXM6+D733HMICQnBiBEjYDAY8PjjjwtZLyIiItVyOvju\n2bMHJSUl+OKLL7By5Uq0bdvW6vO4H9hxy5cvR58+fTBw4EAsXboU169fl7pKsrR7924EBwdDp9Ph\n5MmTUldHdgoLC9GnTx/07NkT69evl7o6sjV37lz4+/sjJCRE6qrI2qVLlxAVFYXg4GCMHj0aO3bs\nkLpKsnTjxg2Eh4djwIABGDZsmP01UGY3+/e//23591/+8hfzCy+84O6XVKycnByz0Wg0G41G8/z5\n880bN26Uukqy9NVXX5m//vpr8+jRo80nTpyQujqyM2DAAHNBQYH5woUL5t69e5vLy8ulrpIsFRYW\nmk+ePGnu16+f1FWRtbKyMvOpU6fMZrPZXF5ebu7evXuDz3W6rbq62mw2m803btwwBwcHm7/55hub\nz3V7eknuB3ZcbGwstFottFotxo4di4KCAqmrJEtBQUHo1auX1NWQpaqqKgBAREQE7r//fsTFxaGo\nqEjiWsnTww8/bHPEjm4LCAjAgAEDAAAdOnRAcHAwvvzyS4lrJU+tWrUCAPz2228wGAxo0aKFzeeK\nktv5+eefR0BAAD777DM8/fTTYryk4r3//vuW7VxEjjp+/DiCgoIsP/ft25e7EEgw58+fR2lpKYYO\nHSp1VWTJZDIhNDQU/v7+WLx4Mbp27WrzuYIE39jYWISEhDT675NPPgEA/O1vf8PFixcxdOhQPPPM\nM0K8pGLZaysAePnll+Hr64vp06dLWFNpOdJORCSea9euYcaMGVizZg1at24tdXVkSavVori4GOfP\nn8fbb7+NU6dO2XyuXogXzM3NtfucVq1aYe7cuR6/KtpeW23ZsgXZ2dk4dOiQSDWSJ0feU9TYkCFD\nsHz5csvPpaWlGDdunIQ1IjWora1FYmIiZs+ejSlTpkhdHdl74IEHMGHCBBQVFSEsLMzqc9w+7Mz9\nwI7LysrCqlWrsH//fs6NO8jseo4YVfHz8wNQt+L5woULyM3NRXh4uMS1IiUzm82YN28e+vXrh6VL\nl0pdHdmqqKjAr7/+CgCorKxETk5Ok19UBMlw1ZRp06bh66+/ho+PD0aPHo3nnnuOixxs6NmzJ2pq\natCuXTsAwPDhw/H2229LXCv52bdvH1JTU1FRUQE/Pz+EhYXh008/lbpaslFQUIBFixahtrYWqamp\nSE1NlbpKspScnIyCggJUVlaiU6dOePnllzFnzhypqyU7n332GSIiItC/f39oNHU5i1999VWOqNyl\npKQEKSkpMBqNCAgIwMyZM/HYY4/ZfL7bgy8RERE1JMpqZyIiIrqNwZeIiEhkDL5EREQiY/AlIiIS\nGYMvERGRyBh8iYiIRPb/AV7ua+oLceA+AAAAAElFTkSuQmCC\n" } ], "prompt_number": 6 }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Testing classification model accuracy with cross validation" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Just like regression models, classification models can suffer from over\n", "fitting. When choosing parameters to include in the model cross\n", "validation is the best method to determine which parameters are essential\n", "for a robust classification model. However rather than R2 we will use\n", "percent correct as our measure of accuracy.\n", "\n", "Let's pretend we collected 2 independent data sets on the above\n", "experiment where we are trying to predict whether patients live or die as\n", "a function of their symptoms. We will fit our classification model on\n", "the first data set and assess how well the model cross validates to the\n", "second data set." ] }, { "cell_type": "code", "collapsed": false, "input": [ "# Define a function to generate fake data\n", "def make_dataset():\n", " X = randn(100, 2)\n", " noise = randn(100)\n", " y = 0.7 * X[:, 0] + 0.5 * X[:, 1] + noise\n", " y = (y > 0).astype(int)\n", " return X, y\n", "\n", "# Make two datasets as above\n", "X1, y1 = make_dataset()\n", "X2, y2 = make_dataset()" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 7 }, { "cell_type": "code", "collapsed": false, "input": [ "# Fit the linear model\n", "line = LogisticRegression().fit(X1, y1)\n", "# Add additional columns where the predictors are squared\n", "X1_quad = hstack((X, X ** 2))\n", "from sklearn.preprocessing import scale\n", "X1_quad = scale(X1_quad)\n", "\n", "# Predict with the line and quad models\n", "line_model = LogisticRegression().fit(X1, y1)\n", "yhat_line1 = line_model.predict(X1)\n", "quad_model = LogisticRegression().fit(X1_quad, y1)\n", "yhat_quad1 = quad_model.predict(X1_quad)\n", "\n", "# Plot the models and estimates\n", "figure(figsize=(12, 5))\n", "# Plot the actual classes of Y1\n", "subplot(121)\n", "plot(X1[y1==0, 0], X1[y1==0, 1], \"o\", color=colors[0], markersize=15)\n", "plot(X1[y1==1, 0], X1[y1==1, 1], \"o\", color=colors[2], markersize=15)\n", "\n", "# Plot the predicted classs from the linear model\n", "plot(X1[yhat_line1==0, 0], X1[yhat_line1==0, 1], \"o\", color=colors[0], markersize=7)\n", "plot(X1[yhat_line1==1, 0], X1[yhat_line1==1, 1], \"o\", color=colors[2], markersize=7)\n", "\n", "# Now do the same for the quadratic model\n", "subplot(122)\n", "plot(X1[y1==0, 0], X1[y1==0, 1], \"o\", color=colors[0], markersize=15)\n", "plot(X1[y1==1, 0], X1[y1==1, 1], \"o\", color=colors[2], markersize=15)\n", "\n", "# Plot the predicted classs from the linear model\n", "plot(X1[yhat_quad1==0, 0], X1[yhat_quad1==0, 1], \"o\", color=colors[0], markersize=7)\n", "plot(X1[yhat_quad1==1, 0], X1[yhat_quad1==1, 1], \"o\", color=colors[2], markersize=7)\n", "\n", "# Compute the accuracy for each model\n", "acc_line = line_model.score(X1, y1)\n", "acc_quad = quad_model.score(X1_quad, y1)\n", "print \"Linear: %.2f\" % acc_line\n", "print \"Quadratic: %.2f\" % acc_quad" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ "Linear: 0.82\n", "Quadratic: 0.60\n" ] }, { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAr8AAAE1CAYAAADu2ojNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nP6Ep4wyRQgweAAAAAElFTkSuQmCC\n" } ], "prompt_number": 8 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Asses how well each model classifies the new dataset (cross validation).\n", "To do this apply the model estimated on dataset 1 to the predictors and\n", "outputs from dataset 2." ] }, { "cell_type": "code", "collapsed": false, "input": [ "line_cv_acc = line_model.score(X2, y2)\n", "X2_quad = scale(hstack((X2, X2 ** 2)))\n", "quad_cv_acc = quad_model.score(X2_quad, y2)\n", "print \"Linear model cross-validated accuracy: %.2f\" % line_cv_acc\n", "print \"Quadratic model cross-validated accuracy: %.2f\" % quad_cv_acc" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ "Linear model cross-validated accuracy: 0.72\n", "Quadratic model cross-validated accuracy: 0.68\n" ] } ], "prompt_number": 9 } ], "metadata": {} } ] }