Edge_attack_rand.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "%pylab inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import networkx as nx\n",
    "from networkx.drawing.nx_agraph import graphviz_layout\n",
    "import random\n",
    "import sklearn as sk\n",
    "from time import time\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from utils import calc_cc, calc_dist, calc_emd, number_of_fake_edges, read_mtx_graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from rand_pert import rand_pertub, link_mirage, smart_link_anon, smart_pertub, calc_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 80 ms, sys: 8 ms, total: 88 ms\n",
      "Wall time: 86.6 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "from_file = True\n",
    "filename = \"data/p2p-Gnutella08/p2p-Gnutella08.mtx\"\n",
    "#filename = \"data/ca-AstroPh/ca-AstroPh.mtx\"\n",
    "filename = \"data/email-Enron/email-Enron.mtx\"\n",
    "filename = \"data/yeast/yeast.mtx\"\n",
    "filename = \"data/USpowerGrid/USpowerGrid.mtx\"\n",
    "filename = \"data/G22/G22.mtx\"\n",
    "\n",
    "if from_file:\n",
    "    g = read_mtx_graph(filename)\n",
    "else:\n",
    "    #g = nx.erdos_renyi_graph(n=1000, p=0.1)\n",
    "    g = nx.powerlaw_cluster_graph(n=4000, m=30, p=0.1)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19990\n",
      "2000\n"
     ]
    }
   ],
   "source": [
    "print(g.number_of_edges())\n",
    "print(g.number_of_nodes())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "118825\n",
      "4000\n"
     ]
    }
   ],
   "source": [
    "print(g.number_of_edges())\n",
    "print(g.number_of_nodes())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split the graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from attack import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "alpha = 0.25"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 176 ms, sys: 4 ms, total: 180 ms\n",
      "Wall time: 167 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "G1, G2 = edge_split(g, alpha)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "M = 10\n",
    "t = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1.72 s, sys: 28 ms, total: 1.75 s\n",
      "Wall time: 1.71 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "f = calc_features(G1)\n",
    "G_aux = smart_link_anon(G1, M, t, 0.5, f)\n",
    "G_aux.add_nodes_from(G1.nodes())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.6 s, sys: 48 ms, total: 3.65 s\n",
      "Wall time: 3.6 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "f = calc_features(G2)\n",
    "G_san = smart_link_anon(G2, M, t, 0.5, f)\n",
    "G_san.add_nodes_from(G2.nodes())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 6197, ' Now edges ', 4650)\n",
      "% of fake edges\n",
      "0.237849462366\n",
      "% of distortion\n",
      "1.1784734549\n",
      " EMD of deg distibution\n",
      "1.81771780325\n",
      " CC diff\n",
      "0.0227549767167\n",
      "CPU times: user 10.6 s, sys: 36 ms, total: 10.7 s\n",
      "Wall time: 10.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 2243, ' Now edges ', 2016)\n",
      "% of fake edges\n",
      "0.362103174603\n",
      "% of distortion\n",
      "1.32545697726\n",
      " EMD of deg distibution\n",
      "0.979989763033\n",
      " CC diff\n",
      "0.133930411092\n",
      "CPU times: user 10.8 s, sys: 8 ms, total: 10.8 s\n",
      "Wall time: 10.8 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 2243, ' Now edges ', 1886)\n",
      "% of fake edges\n",
      "0.334040296925\n",
      "% of distortion\n",
      "1.28087382969\n",
      " EMD of deg distibution\n",
      "0.97740811879\n",
      " CC diff\n",
      "0.121664774632\n",
      "CPU times: user 10 s, sys: 8 ms, total: 10 s\n",
      "Wall time: 10 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 2243, ' Now edges ', 1902)\n",
      "% of fake edges\n",
      "0.330178759201\n",
      "% of distortion\n",
      "1.27998216674\n",
      " EMD of deg distibution\n",
      "0.947993460668\n",
      " CC diff\n",
      "0.134331644249\n",
      "CPU times: user 10.8 s, sys: 0 ns, total: 10.8 s\n",
      "Wall time: 10.8 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 2243, ' Now edges ', 1853)\n",
      "% of fake edges\n",
      "0.332973556395\n",
      "% of distortion\n",
      "1.27507802051\n",
      " EMD of deg distibution\n",
      "0.96010640233\n",
      " CC diff\n",
      "0.106435162466\n",
      "CPU times: user 11.8 s, sys: 4 ms, total: 11.8 s\n",
      "Wall time: 11.8 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 2243, ' Now edges ', 1893)\n",
      "% of fake edges\n",
      "0.340729001585\n",
      "% of distortion\n",
      "1.28756130183\n",
      " EMD of deg distibution\n",
      "0.952857519299\n",
      " CC diff\n",
      "0.13228939233\n",
      "CPU times: user 11.5 s, sys: 16 ms, total: 11.5 s\n",
      "Wall time: 11.5 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 3069, ' Now edges ', 2580)\n",
      "% of fake edges\n",
      "0.320542635659\n",
      "% of distortion\n",
      "1.26946888237\n",
      " EMD of deg distibution\n",
      "1.04139652898\n",
      " CC diff\n",
      "0.11561717201\n",
      "CPU times: user 7.71 s, sys: 16 ms, total: 7.72 s\n",
      "Wall time: 7.72 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G2.number_of_edges(), \" Now edges \", G_san.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G2, G_san))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G2, G_san))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G2, G_san))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G2, G_san))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.04 s, sys: 32 ms, total: 4.08 s\n",
      "Wall time: 4.01 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "cache = {}\n",
    "f = calc_features(G1)\n",
    "G_aux = smart_pertub(G1, M, t, 0.5, f, cache)\n",
    "G_aux.add_nodes_from(G1.nodes())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 7.72 s, sys: 64 ms, total: 7.79 s\n",
      "Wall time: 7.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "cache = {}\n",
    "f = calc_features(G2)\n",
    "G_san = smart_pertub(G2, M, t, 0.5, f, cache)\n",
    "G_san.add_nodes_from(G2.nodes())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 6197, ' Now edges ', 6278)\n",
      "% of fake edges\n",
      "0.714877349474\n",
      "% of distortion\n",
      "1.72422139745\n",
      " EMD of deg distibution\n",
      "1.38847811042\n",
      " CC diff\n",
      "0.0141698462152\n",
      "CPU times: user 14 s, sys: 16 ms, total: 14 s\n",
      "Wall time: 14 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 2243, ' Now edges ', 2148)\n",
      "% of fake edges\n",
      "0.515363128492\n",
      "% of distortion\n",
      "1.4935354436\n",
      " EMD of deg distibution\n",
      "0.822551895232\n",
      " CC diff\n",
      "0.120491682453\n",
      "CPU times: user 5.99 s, sys: 16 ms, total: 6.01 s\n",
      "Wall time: 6 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 6516, ' Now edges ', 6733)\n",
      "% of fake edges\n",
      "0.562453586811\n",
      "% of distortion\n",
      "1.58118477594\n",
      " EMD of deg distibution\n",
      "1.0420262383\n",
      " CC diff\n",
      "0.0350216652959\n",
      "('Originial edges ', 8995, ' Now edges ', 9339)\n",
      "% of fake edges\n",
      "0.583145947104\n",
      "% of distortion\n",
      "1.60544747082\n",
      " EMD of deg distibution\n",
      "1.04839306098\n",
      " CC diff\n",
      "0.0245119381425\n",
      "CPU times: user 3min 35s, sys: 84 ms, total: 3min 35s\n",
      "Wall time: 3min 35s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G2.number_of_edges(), \" Now edges \", G_san.number_of_edges())\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G2, G_san))\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G2, G_san))\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G2, G_san))\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G2, G_san))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1.26 s, sys: 32 ms, total: 1.29 s\n",
      "Wall time: 1.24 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "G_aux = link_mirage(G1, M, t, 0.5)\n",
    "G_aux.add_nodes_from(G1.nodes())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1.43 s, sys: 52 ms, total: 1.48 s\n",
      "Wall time: 1.36 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "G_san = link_mirage(G2, M, t, 0.5)\n",
    "G_san.add_nodes_from(G2.nodes())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 6197, ' Now edges ', 3297)\n",
      "% of fake edges\n",
      "0.417955717319\n",
      "0.488121546961\n",
      "% of distortion\n",
      "1.2223656608\n",
      "1.2070784015\n",
      " EMD of deg distibution\n",
      "3.0973454604\n",
      "4.9500203253\n",
      " CC diff\n",
      "0.0264922796722\n",
      "0.0230289206005\n",
      "CPU times: user 7.25 s, sys: 36 ms, total: 7.29 s\n",
      "Wall time: 7.26 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(number_of_fake_edges(G2, G_san))\n",
    "\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(calc_dist(G2, G_san))\n",
    "\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(calc_emd(G2, G_san))\n",
    "\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))\n",
    "print(calc_cc(G2, G_san))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Originial edges ', 2051, ' Now edges ', 2279)\n",
      "% of fake edges\n",
      "0.305397103993\n",
      "0.298312639287\n",
      "% of distortion\n",
      "1.33934666017\n",
      "1.33404634581\n",
      " EMD of deg distibution\n",
      "0.500534918855\n",
      "0.464229931432\n",
      " CC diff\n",
      "0.0312969009566\n",
      "0.0237023930686\n",
      "CPU times: user 51.9 s, sys: 80 ms, total: 52 s\n",
      "Wall time: 52 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Originial edges \", G1.number_of_edges(), \" Now edges \", G_aux.number_of_edges())\n",
    "\n",
    "print(\"% of fake edges\")\n",
    "print(number_of_fake_edges(G1, G_aux))\n",
    "print(number_of_fake_edges(G2, G_san))\n",
    "\n",
    "print(\"% of distortion\")\n",
    "print(calc_dist(G1, G_aux))\n",
    "print(calc_dist(G2, G_san))\n",
    "\n",
    "print(\" EMD of deg distibution\")\n",
    "print(calc_emd(G1, G_aux))\n",
    "print(calc_emd(G2, G_san))\n",
    "\n",
    "print(\" CC diff\")\n",
    "print(calc_cc(G1, G_aux))\n",
    "print(calc_cc(G2, G_san))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Start adding false examples\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Aux finished 21.0361759663 sec\n",
      "Start adding positive examples\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Start adding false examples\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "San finished 34.5330729485 sec\n"
     ]
    }
   ],
   "source": [
    "X_naive, Y_naive = gen_trainset(G1, G2, alpha, \"data/train_naive.dump\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.0997821350763\n",
      "Finished : 0.199564270153\n",
      "Finished : 0.299346405229\n",
      "Finished : 0.399128540305\n",
      "Finished : 0.498910675381\n",
      "Finished : 0.598692810458\n",
      "Finished : 0.698474945534\n",
      "Finished : 0.79825708061\n",
      "Finished : 0.898039215686\n",
      "Finished : 0.997821350763\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0994152046784\n",
      "Finished : 0.198830409357\n",
      "Finished : 0.298245614035\n",
      "Finished : 0.397660818713\n",
      "Finished : 0.497076023392\n",
      "Finished : 0.59649122807\n",
      "Finished : 0.695906432749\n",
      "Finished : 0.795321637427\n",
      "Finished : 0.894736842105\n",
      "Finished : 0.994152046784\n",
      "Aux finished 5.66770291328 sec\n",
      "Start adding positive examples\n",
      "Finished : 0.0999687597626\n",
      "Finished : 0.199937519525\n",
      "Finished : 0.299906279288\n",
      "Finished : 0.39987503905\n",
      "Finished : 0.499843798813\n",
      "Finished : 0.599812558575\n",
      "Finished : 0.699781318338\n",
      "Finished : 0.799750078101\n",
      "Finished : 0.899718837863\n",
      "Finished : 0.999687597626\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0996818663839\n",
      "Finished : 0.199363732768\n",
      "Finished : 0.299045599152\n",
      "Finished : 0.398727465536\n",
      "Finished : 0.498409331919\n",
      "Finished : 0.598091198303\n",
      "Finished : 0.697773064687\n",
      "Finished : 0.797454931071\n",
      "Finished : 0.897136797455\n",
      "Finished : 0.996818663839\n",
      "San finished 9.32091903687 sec\n"
     ]
    }
   ],
   "source": [
    "X, Y = gen_trainset(G_aux, G_san, alpha, \"data/train.dump\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.72 s, sys: 72 ms, total: 3.8 s\n",
      "Wall time: 1.87 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest.fit(X, Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.0997495826377\n",
      "Finished : 0.199499165275\n",
      "Finished : 0.299248747913\n",
      "Finished : 0.398998330551\n",
      "Finished : 0.498747913189\n",
      "Finished : 0.598497495826\n",
      "Finished : 0.698247078464\n",
      "Finished : 0.797996661102\n",
      "Finished : 0.89774624374\n",
      "Finished : 0.997495826377\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0997495826377\n",
      "Finished : 0.199499165275\n",
      "Finished : 0.299248747913\n",
      "Finished : 0.398998330551\n",
      "Finished : 0.498747913189\n",
      "Finished : 0.598497495826\n",
      "Finished : 0.698247078464\n",
      "Finished : 0.797996661102\n",
      "Finished : 0.89774624374\n",
      "Finished : 0.997495826377\n",
      "Test set generated in  34.0687160492 sec\n"
     ]
    }
   ],
   "source": [
    "X_t, Y_t = gen_testset(G_aux, G_san, \"data/test_test.dump\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Start adding false examples\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Finished : 0\n",
      "Test set generated in  299.437201023 sec\n"
     ]
    }
   ],
   "source": [
    "X_t_naive, Y_t_naive = gen_testset(G1, G2, \"data/test_test_naive.dump\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Anonymization\n",
      "0.52027027027\n",
      "Naive:\n",
      "0.866073730343\n",
      "CPU times: user 9.91 s, sys: 212 ms, total: 10.1 s\n",
      "Wall time: 4.79 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "print(\"Naive:\")\n",
    "forest.fit(X_naive, Y_naive)\n",
    "print(forest.score(X_t_naive, Y_t_naive))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "y_score = forest.predict_proba(X_t_naive)\n",
    "y_true = Y_t_naive\n",
    "fpr, tpr, thresholds = roc_curve(y_true, y_score[:, 1])\n",
    "roc_auc = auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIpKQEoSIiCSkBCEiIgn9P5zvAtUDNAtwAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f51b2f01d90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, 'b',\n",
    "label='GS, AUC = %0.2f'% roc_auc)\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.1,1.2])\n",
    "plt.ylim([-0.1,1.2])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Anonymization\n",
      "0.510135135135\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest.fit(X, Y)\n",
    "print(\"Anonymization\")\n",
    "print(forest.score(X_t, Y_t))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Anonymization\n",
      "0.626257545272\n",
      "CPU times: user 5.7 s, sys: 120 ms, total: 5.82 s\n",
      "Wall time: 2.38 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest.fit(X, Y)\n",
    "print(\"Anonymization\")\n",
    "print(forest.score(X_t, Y_t))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Anonymization\n",
      "0.651633349586\n",
      "CPU times: user 6.73 s, sys: 136 ms, total: 6.86 s\n",
      "Wall time: 2.83 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest.fit(X, Y)\n",
    "print(\"Anonymization\")\n",
    "print(forest.score(X_t, Y_t))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Anonymization\n",
      "0.528589315526\n",
      "CPU times: user 4.14 s, sys: 112 ms, total: 4.25 s\n",
      "Wall time: 2.05 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Smart link 0.5\n",
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest.fit(X, Y)\n",
    "print(\"Anonymization\")\n",
    "print(forest.score(X_t, Y_t))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "y_score = forest.predict_proba(X_t)\n",
    "y_true = Y_t\n",
    "fpr, tpr, thresholds = roc_curve(y_true, y_score[:, 1])\n",
    "roc_auc = auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f51b2d9b850>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, 'b',\n",
    "label='GS, AUC = %0.2f'% roc_auc)\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.1,1.2])\n",
    "plt.ylim([-0.1,1.2])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npJQgREQkKSUIERFJSglCRESS+v/4956LWS1+SQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f69617fe710>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Smart pertubation\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, 'b',\n",
    "label='GS, AUC = %0.2f'% roc_auc)\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.1,1.2])\n",
    "plt.ylim([-0.1,1.2])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XecFPX5wPHPcwgYOcADC2BBiqixoGgQK6cYe6+AIpYY\nSwQlmh9GjIAxxoJiFEGjKC2EZu+FcEgVpBkREGkCIhcO8EClHPf8/vjO4bLM3u3d7exsed6v175u\nd2Z25pm7vXn2W+b7FVXFGGOMiZYTdgDGGGNSkyUIY4wxvixBGGOM8WUJwhhjjC9LEMYYY3xZgjDG\nGOPLEoSpkIh0FpEPwo4jlYjIJhE5JITjNhWRUhHJiP9dEflSRE6vwvvsM5kEGfEhyyYislxEfhKR\nYhH5TkReEZG9gjymqo5U1XODPEYkETlZRMZ757hBRN4UkSOSdXyfeCaIyE2Ry1S1rqouD+h4rURk\njIj8zzv/uSLSQ0Sk7PBBHLeyvETVvDr7UNWjVPXTCo6zW1JM9mcyW1mCSD8KXKCq9YBjgeOAP4cb\nUtWISA2fZScBHwKvA42BZsAXwJQgvrH7xRAmEWkBTAdWAEepah5wFXA8UDfBx6ruuVc5UVXy2OId\nSyra0CSYqtojjR7AMuDMiNePAW9HvK4F9MNdYNYAA4HaEesvAeYAPwCLgbO95fWAl4DvgJXAXwHx\n1nUFJnnPBwFPRMX0BnC397wxMA4oBJYA3SK26w2MBYYDG4GbfM7vU+BZn+XvAUO85+29GP8M/A9Y\nCnSO53cQ8d7/89YNBfYG3vZiLvKeN/G2fxgoAX4CioFnvOWlQHPv+SvAAOAdb5tpQLOIeM4GFgIb\ngOeAAr9z97YdHvn39Fnf1Dv29d75FQL3R6z/DTDVO9Zq4Flgj4j1pcAdwNfAEm/Z08C33mdiJnBq\nxPY5wP3AN965zQQOBCZ6+9rsLb/K2/5C3OdrAzAZODrqs/t/wDzgZ6AGEZ9nL/aZXhxrgH7e8hXA\nDmCTd6wTifhMetscCXzk/f3WAPeF/b+aCY/QA7BHJf9gu/5DHYj7dv1UxPqncRfs+kAd4E3gb966\ntrgLc9n7GwOtvOdv4C6kewL74L7F3uKt6wp86j0/DVgRcby9cRfP/XHf8D4Henn//Id4F5bfetv2\nBrYCF3mva0ed269wF+P2Pud9A7Dae94e2A48AdQETvcuVIfG8Tsoe+8j3ntrAw2Ay7zndYDRwOsR\nx55A1AXdu2BFJoh1uG/5OcAIYKS3rqF3wbvEW9fd+x3EShBrgK7l/P3LEsQLuER4DLAFOMxb38b7\nOwtwMDAf6B7x/lJcCa0+vyTNzt7fMQfo4cVQy1v3J9wFvaX3+mggL2JfkYmwDbAWOME7fhfc57Vm\nxGd3NtAk4tiRn+epwLXe872AthHnvAPvC4vPZzIX98Xmbu93Ugf4Tdj/q5nwCD0Ae1TyD+b+oYq9\nRynwMVAvYv3mqH/ak4Cl3vPngSd99rmfd5GJLGl0BP7jPd/5z+i9Xo73LRP4HfCJ9/xEYHnUvu8D\nBnvPewMF5ZzbAd45tfJZdw6w1XveHtgG7BmxfjTQK47fQXvvXGuWE8exQFHEa78EEV2C+GfEuvOA\nr7znXYApUe/9Nnp/Eeu24ZXqYqwvu1g2jlj2GXB1jO3vAl6Nirt9BZ+x9Xjf/HElnwtjbLfzd+C9\nHgj0jdpmIXBaxGe3q8/nuSxBFHifkYYxzjknYllkgugIzArqfy6bH3tg0tElqjpBRE4DRuK+8ReL\nyL64b16zfmnPJIdf6m4PAt712V9T3LfpNd77xHt8G+P4o4FOuCqEzrhqEXDfWA8QkfXea/GOH9kI\nubKc89qAu+g0xlWBRGqM+5a+c1tV3RLxegXQJI7fAcD/VHV72QsR+RWu1HEO7pu0ALkiIupdgeLw\nfcTzn3DfasF9W44+51Xl7KcId64VWet3PBE5FHgK9y3+V8AewKzyji8i9wA3Rxy3Lu4zBe4zszSO\neMB9jq4XkW5lu8Z9rprEOnaUm3FVmwtFZCnwkKr6fV6jHYSrzjQJZo3U6UkAVHUSrg79SW/5OtzF\n4khVbeA99lbV+t76lUALn/2txH2rbui9J8973zExjv9v4EoRORhXang1Yj9LI46dp6r1VfWiiPfG\nvOCq6k+4+vurfFZfDXwS8TrPu7CXORhXzVDR78AvhnuAQ3HVEnvjqqzgl6QSb5LwswZ3AYt0YDnb\nfwJcUY3jDQIWAC28c+nF7o27O89HRE7FtQtc6f298nCl07L3xPrM+FmJq8qL/Pvnqupov2NHU9Ul\nqtpZVfcFHgfGeX/jin7/K4GWccZoKsESRPp7GvitiBzjfdt9EXja+yaNiBwgImd72w4GbhSRM8Rp\nIiKHqer3uAa+/iJS11vXPFb/dFWdi7sQvwR8oKrF3qoZuJLM/4nIniJSQ0SOFJETKnE+9wFdReRO\nEckVkTwReRhoB/SN2E6AviJS0ytJXQCMieN34KcurtG0WEQaAH2i1q8Fqtqd813gKBG52Pt93Ilr\nr4mlN3CyiDwmIvt78bcUkeEiUs/bprzePHWBYlX9SUQOB26vIL66uDaZIhGpJSIPsmtvqZeAv4pI\nSy+Wo0Ukz1v3Pbv+Xl4EbhORtt62dUTkfBGpU0EMeNtfKyJlJZcfcIlhB64jQimxE9U7wP4i0t07\nh9yyGEz1WIJIP7t8m1LVdbhSxF+8RffhGoani8hG3IW/lbftTOBGXFL5AVfne7D3vutxDXxf4eqg\nxwKNyonj30AH4F8RsZQCF+Hq8Jfheti8iOshFd/JqU7BVfVcgfv2vQxoDZyiqpFVHWtwVVLf4aq4\nblXVxd66nrF+BzE8jauWWodrKH0vav0/gKtEpEhEni4LNc7zKcKViJ7w9n84riF/a4ztl+LaTJoB\n80VkA+5vMRPXi8fv2JGv7wWuFZFiXEP2qHK2Bddg/QGuSm8ZrvQVWSX2FDAG+EhEfsAljLKSW19g\nmIisF5ErVXUWcAswwKtm/BrXVhDr2NHLzvXOuRjoD1yjqttU9Wfgb7iuzuujL/6quhn4LXAxLml9\nDeT7HMtUUlk3RmPShoi0B4ar6sEVbpxivJvdVuG65U4MOx5jymMlCGMCJiJni0h9EamNaxMA143Y\nmJRmCcKY4J2E62VTiGsruURVfauYjEklVsVkjDHGl5UgjDHG+EqbG+VExIo6xhhTBapapYEO06oE\nkYxby3v37h367e12Lpl/Ppl0Lpl2Ppl0LqrV+16dVgnCGGNM8liCMMYY48sSRJT8/PywQ0iYTDoX\nyKzzyaRzgcw6n0w6l+pKm26ulRtY0xhjDICIoNnQSG2MMSZ5LEEYY4zxZQnCGGOML0sQxhhjfFmC\nMMYY48sShDHGGF+WIIwxxviyBGGMMcaXJQhjjDG+LEEYY4zxZQnCGGOML0sQxhhjfFmCMMYY48sS\nhDHGGF+WIIwxxviyBGGMMcaXJQhjjDG+LEEYY4zxZQnCGGOML0sQxhhjfFmCMMYY48sShDHGGF+W\nIIwxxviyBGGMMcZXoAlCRAaLyFoR+aKcbZ4RkcUiMldEjg0yHmOMMfELugTxCnBOrJUich7QQlUP\nBW4Fng84HmOMMXEKNEGo6mRgQzmbXAIM87b9DKgvIvsHGZMxxpj47BHy8Q8AVka8Xu0tWxtOOMaY\nylCFH3+E4mLYtOmXn9HPS0rctmXvSeWfItC/f+V+D5kq7AQhPss01sZ9+vTZ+Tw/P5/8/PzER2RM\nFtuyBb77DpYuhYULYdEiWLsWfvrJJYIff9w1Afz4I+y5J9SrB3Xr/vIz+nnNmm7/IhX/LHvEu30Q\nP9NZQUEBBQUFCdmXqMa8HifmACJNgbdV9Rifdc8DE1R1tPd6IdBeVXcrQYiIBh2rMdmgpAQWL4Yv\nv3RJYMEC93PFCti8GRo1gmbN4LDD3KNJE6hTxz322mvXC39uLuwR9tfM6ioshGXL4MQTw44kECKC\nqlYp9SXjTyv4lxQA3gL+AIwWkXbARr/kYIypmg0b4IsvYN68Xx4LFriL/lFHwRFHwHnnwd13u6TQ\nsCHkZFPn9zFjoHt36NEjYxNEdQRaghCRkUA+0BDXrtAbqAWoqv7T22YAcC7wI3Cjqs6OsS8rQRhT\ngZIS+PhjGDECJk1yCeLoo6F1618eRx/tvvlntcJC+MMfXDFqyJCMTg7VKUEEXsWUKJYgjIntu+9c\nw+rw4dC0KXTp4koGzZplWYkgHu++CzffDF27Qt++rhElg6V6FZMxJiDFxfDXv8Lgwe56N3Giazcw\n5cjLgzffzOhSQ6JYgjAmzZSWwrRp8MYb8K9/wbnnwvz50Lhx2JGliZNPDjuCtGEJwpg0oAqzZsGo\nUTB6NNSvD1dcAR9+6NoUjAmCJQhjUlhJCYwbB336wI4d0LEjvP++64FkKjBmjOvP26tX2JGkLUsQ\nxqSY7duhoMAlhtdfh0MPheeegzPPzIwbuQIX3UPJVJn1bzAmRaxZA3fd5e5ReOABaNkSPvsMpkyB\nDh0sOcRlzBg45hho3hzmzLGG6GqyEoQxISsshMcec192b7gBPv/cdVU1lfT00/DCC9ZDKYHsPghj\nQrJ+PfTr565pnTvDn//sSg+mijZudPc0ZPh9DZVVnfsgrIrJmCT74QfX6NyqFaxb52pCnn3WkkO1\n7b23JYcEswRhTJJs3OhuamvZEpYvhxkz4J//hIMPDjuyNLRpU9gRZAVLEMYErKjIlRhatoQlS1yj\n85Ahrh3VVFJhIVx1FdxxR9iRZAVLEMYEZOFCuOUWlxhWroTp011iaNUq7MjSVGQPpRdfDDuarGC9\nmIxJsOXLXYnhvffgzjvdpDv77Rd2VGks8r4G66GUVJYgjEmQ+fPhiSfgrbdcYli82A2JYarpjTdc\nqWH4cGuETjLr5mpMNaxe7aqNxo2D7793ieGOO9yAocakApsPwpgkW7wYHnnE1Xh07AjXXAOnngo1\naoQdmTG7svsgjEmiESPciNHNm8M338DAgdC+vSWHaissdINQmZRhCcKYOC1a5HpYPvQQjB8Pf/kL\nNGgQdlQZQNWNYX7MMTB5ctjRmAjWSG1MBUpL3Q1uAwbAvffC0KGw115hR5UhCgtdo81XX1kPpRRk\nCcKYcqxcCbfd5qb2/OILm7Utod5/H2680Y1QOGKE9VBKQVbFZIyPbdvg0UfhuOPgN79xVUqWHBKs\ncWNXanj0UUsOKcpKEMZE+eQT1121bD6GFi3CjihDHXts2BGYCliCMMazahXcc48bRO+ZZ+Cii8KO\nyJhwWRWTyXrLl0O3bq4TzWGHuTuiLTkkSFkPpZ49w47EVIGVIEzWmjsXHn8cPvzQDao3f761MyRU\nZA+lV14JOxpTBYGXIETkXBFZKCJfi8huXyNE5CAR+Y+IzBaRuSJyXtAxmexWVAQXXwwXXOAaoZct\nc+2klhwSJPK+hpYtYfZs676apgItQYhIDjAA6AB8B8wUkTdVdWHEZg8Ao1X1BRE5AngPaBZkXCZ7\nzZsHV1wBl18OY8dC7dphR5SB/vlP+Mc/7L6GDBB0CaItsFhVV6jqdmAUcEnUNqVAPe/53sDqgGMy\nWej77+H3v4ff/hYefNBVLVlyCMi111qpIUPElSBEpJaItKzC/g8AVka8XuUti9QX6CIiK4F3gG5V\nOI4xvkpKXI+ko46CunXdcBnXXx92VBkuN9fua8gQFVYxicgFwFNALaCZiBwL9FbVy+LYv98IgtFD\nsnYCXlHV/iLSDhgBHOm3sz59+ux8np+fT35+fhwhmGw1Y4a7C7p+fZg0CY44IuyIMoyqm2jbxjZP\nKQUFBRQkaNDDCof7FpFZuDaECap6nLfsv6p6dIU7dxf8Pqp6rvf6PkBV9bGIbb4EzlHV1d7rJcCJ\nqroual823LeJy7ffurGT3nnHVSVddx1IlQY7NjGV9VACNxmGSVlBD/e9XVU3Ri2L90o9E2gpIk1F\npBbQEXgrapsVwFkAXiN17ejkYEw8Skqgd2/XM2nffV3vyi5dLDkkVHQPpREjwo7IBCieXkwLRORq\nIEdEmgF3AdPj2bmq7hCRO4GPcMlosKouEJG+wExVfQe4F3hRRHrgGqy7VuVETHZbvty1jdap46Yu\nti6rAbCRV7NOPFVMdYAHgbO9RR8CfVX154Bji47DqpiMr9Gj3Z3Q//d/8Mc/Qo6NDxCMUaPc3YV9\n+lgjdBoJdMpREblcVV+raFnQLEGYaJs3Q/fubo6Zf/8bjj8+7IiMST1Bt0E84LOsV1UOZkyizJ7t\nEoLqL8+NMYkVsw1CRM4BzgUOEJGnIlbVw7UVGJN0paXw9NPw97+7+xs6dQo7ogxUWAiffw7nnx92\nJCZk5ZUgCoEvgS3A/IjHR4CNl2SS7qef4JJL3BAZM2ZYcki4yB5KM2eGHY1JAfG0QeypqluSFE95\ncVgbRBbbtMkNwX3QQfDyy1CzZtgRZZjokVeth1LGCLoN4gARGSUiX3gjsn4tIl9X5WDGVMXGjXD2\n2dCqFQwdaskh4T75xEZeNb7iuQ9iCPAw0A9XtXQj8d8oZ0y1FBW55HDKKW6AULvpLQBNm9p9DcZX\nXENtqOrxkcNriMjnqnpCUiL8JQ6rYsoya9e60VfPO8/N12DJwZjKq04VUzwliK0iIsASEbkNNxx3\n3aoczJh4rV4NZ50F11zjhs+w5GBM8sXTBtEDyAW6A6cAtwA3BRmUyW7ffgvt20PXru6mXUsOCVDW\nQ+m228KOxKSRCquYfN8kcqCqrgognvKOaVVMWWDJEldyuOsuuPvusKPJENZDKasF1otJRH4jIpeK\nyD7e6yNFZBhxDtZnTGVMnQqnnw49e1pySAibG9pUU8wShIj8HbgCmIebI/od4A7gMWCQqv6UrCC9\neKwEkcGGDHGJYcgQ1yhtEmD4cHfLuZUaslogg/WJyFfA8ar6s4g0wE0derSqLq16qFVnCSJzvfqq\nKzF88gkcdljY0WSQLd79rTbyalYLqhfTlrIhvVV1vYh8HVZyMJlrxgy4/Xb44ANLDglnicFUU3kJ\normIlA3pLbj5qHcO8a2qlwcamcl4S5bApZfC4MHQpk3Y0aQxVVi3zk2jZ0wClVfF1KG8N6rq+EAi\nisGqmDLL2rXu7ug//QluvTXsaNJYWQ+ln36C994LOxqTggKdMChVWILIHMXFkJ/vRmbt3TvsaNKU\nKowZ4/oD33CDzfJmYrIEYdLG1q1wwQWu1+WgQXYTXJXYfQ2mEoIeasOYhCgpcXdH168Pzz1nyaHK\nPv/cZdgRI6zUYAIVdwlCRGqr6taA4ynv+FaCSGNbt0Lnzq566e237bpmTLIEOh+EiLQVkf8Ci73X\nrUXk2aoczGSnzZvhwgvd83feseRgTLqIZ7C+Z4ALgSIAVZ0HnBFkUCZzbN8OV14JTZq4UR9q1w47\nojRSWAjjxoUdhcli8SSIHFVdEbVsRxDBmMyi6tpSRdy9DntYi1d8IsdQmjcv7GhMFovnX3aliLQF\nVERqAN0Am3LUVOiRR2DWLJg40ZJD3CJ7KNksbyZk8ZQgbgf+CBwMrAXaecviIiLnishCby7rnjG2\nuVpE5ovIf0VkRLz7NqlJFR580M0f/e67UNeml4pPQYGNvGpSSjxTjjZQ1fVV2rlIDq600QH4DpgJ\ndFTVhRHbtARGA2eoarGI7KOq63z2Zb2Y0sCOHfCHP7iemO+9B/vtF3ZEaWTVKjeVniUGk0BB3wcx\nU0QW4S7ir6nqpkrsvy2wuKwNQ0RGAZcACyO2uQV4TlWLAfySg0kPqm7CsqVLYcIEKzlU2oEHuocx\nKaLCKiZVbQE8DBwP/FdE3hCRjnHu/wDcMOFlVnnLIrUCDhORySIyVUTOiXPfJsU88wx89pmrOrfk\nYEz6i6cNAlWdqqrdgTZAMfCvOPfvV6yJrifaA2gJnA50Bl4SkXpx7t+kiI8+gkcfhbfegtzcsKNJ\nYWU9lDp3DjsSYypUYRWTiOTiqoU6AkcAbwInx7n/VbjG7TIH4toioreZpqqlwHKvOutQYFb0zvr0\n6bPzeX5+Pvn5+XGGYYI0aRJcd52b+OeQQ8KOJoVFj6FkTAAKCgooKChIyL7iaaReDrwNjFHVSZXa\nuesWuwjXSL0GmAF0UtUFEduc4y27wZv7ehZwrKpuiNqXNVKnoAkT4Oqr4d//hrPOCjuaFGUjr5oQ\nBd1I3dz7dl9pqrpDRO4EPsJVZw1W1QUi0heYqarvqOqHInK2iMwHSoB7o5ODSU3jx0PHjjB2rBu+\n28Tw+uvQt6/d12DSTnkTBj2pqveIyOvs3m6Q9BnlrASRWr76yiWFsWOhffuwo0lxJSXuYaUGE4Kg\nShCjvZ8DqrJjk7mKiuDii+GJJyw5xGWPPexWcpOWYvZiUtUZ3tMjVHV85APXWG2y0Pbtrs3hssvc\n3A4mgip8F90Hw5j0FU8315t8lt2c6EBMerj7bldT8uijYUeSYgoL4aqrXPdVqwo1GSJmuVdErsF1\nbW0mIq9FrKoLbAw6MJN6Bg1yvZamTYMaNcKOJkVE91AaMcKmyjMZo7yK0Rm4OSAOBJ6LWL4JmBNk\nUCb1fPyx64gzebKbMtRgI6+ajBczQajqMmAZ8EnywjGpaNIkV3Py2mtuoFHjWbzY5oY2Ga28bq4T\nVbW9iGxg126uAqiqNkhGgBHxWDfXEMyY4aYLHTnSboQzJh1Vp5treQkiR1VLvbuhd6OqSZ1VzhJE\n8s2fD2ee6WaDK5tT2hiTXqqTIMrr5lp29/RBQA0vIZwE3ArUqcrBTPrYssXdJf3II5YcKCx0sx8Z\nk2Xi6eb6Bm660RbAK7iB9EYGGpUJ3X33weGHw01+nZyzReTc0IsWWfdVk3Xiub2zVFW3i8jlwLOq\n+oyIWC+mDPbhh65Beu7cLO6xaT2UjImrBFEiIlcBXYB3vGU1gwvJhOmbb+D6613HnAZJ7YaQQqZO\ntbmhjSG+4b6PAu4ApqrqCBFpBnRW1b8lI8CIOKyROmDFxXDSSXDnnXD77WFHE6KiIpcpLTGYDBBI\nL6aoA5TN+gbwjaqWVOVg1WEJIlilpXDppdCkCTz/fNjRGGMSJdD5IETkNGA4sBp3D0QjEemiqlOq\nckCTmh58EDZuhHHjwo7EGJMq4mmD6A+cr6qnqOrJwAXAP4INyyTT6NGuzWHcOKhVK+xokqSsh9KF\nF1rvJGNiiKcXUy1V/arshTcjXLZcRjLevHmuzeHjj2G//cKOJknKeijNnw9DhmRxVy1jyhdPCWK2\niLwgIqd6j0HYYH0ZYcsWN8bSU0/BsceGHU2SjBnjeii1aAFz5lhDtDHliKcX055Ad+BUXBvEp7j7\nIbYEH94ucVgjdYL98Y+wapWracmKL9EffugmtBgyxBKDyRqB9WISkaOBFsB8VV1cxfgSwhJEYk2Y\nANddB198AQ0bhh1NkqjC1q028qrJKoGMxSQi9+OG2bgW+FhEsnnQhYzy3XfQpQu8/HIWJQdwxSRL\nDsbErbzRXOcDbVX1RxHZF3hPVX+T1Oh2jcdKEAmwdSvk57vOO716hR1NgFasgKZNw47CmNAFUoIA\ntqrqjwCq+r8KtjVpQNX1WDrgALj//rCjCUjZ3NDXXGPdV42ppvK6uTaPmItagBaRc1Or6uWBRmYS\n7oUX3HzS06dnaKP0mDHQvTt07QrDh2foSRqTPOVVMXUo742qOj6QiGKwKqbqmTwZrrgCpkzJwGlD\nCwvhD3+AL7+EV16Bdu3CjsiYlBHIUBvJTgAmOKtWuRqXoUMzMDkArFkDzZu7UoM1QhuTMHEN1let\nA4icCzyNa8MYrKqPxdjuSmAMcIKqzvZZbyWIKti2DU4/3Q3Ed999YUdjjEm2wEdzrSoRyQG+BjoA\n3wEzgY6qujBqu1zgXdw8E3dagkice+91k6G99ZZVyRuTjYLqxRR9kNpV2H9bYLGqrlDV7cAo4BKf\n7f4KPAZsrcIxTAzvvANjx2bQcEOFhTBwYNhRGJM1KkwQItJWRP4LLPZetxaRZ+Pc/wHAyojXq7xl\nkfs/FjhQVd+Lc58mDqtWwe9+ByNHZsjNcGVjKK1Y4SavMMYELp7RXJ8BLsTdVY2qzhORM+Lcv9/3\n1p31RCIiuOHEu1bwHgD69Omz83l+fj75+flxhpFdSkqgUyc37NApp4QdTTVF9lB64w3roWRMBQoK\nCigoKEjIvuIZrG+GqrYVkTmqepy3bJ6qtq5w5yLtgD6qeq73+j5AyxqqRaQe8A2wGW8yIqAIuDi6\nHcLaIOLXqxfMmgXvvQc56Xx74+efu1u+u3aFvn2th5IxVRDojHLAShFpC6iI1AC64Rqe4zETaCki\nTYE1QEegU9lKVS0Gds5CICITgD+qqg0nXkUffui6s86enebJAeDww13retu2YUdiTFaK5xJyO/BH\n4GBgLdDOW1YhVd0B3Al8BMwHRnkTDvUVkQv93kI5VUymfEuXwvXXu3aHjJj8JzfXkoMxIQr8PohE\nsSqm8v34I5x0Evz+9268pbSjmiFdrYxJLYHeByEiLxLRsFxGVX9flQNWlSWI2EpLXaP0Xnu5IbzT\n7jo7Zozrvjp+PNSoEXY0xmSUoNsgPol4vidwGbt2XTUh2r4dbroJvv/etT+kVXKIHkPJkoMxKaXS\nVUze3dGTVfXkYEKKeVwrQUT5+We4+mpXOzNmjCtBpI3IkVeth5IxgQm6BBGtGbB/VQ5mEuumm1w7\n7rBhULNm2NFUwpQp0Lu33ddgTIqLpw1iA7+0QeQA64H7VHVMwLFFx2EliAgTJ7oeSwsWpFnJocy2\nbVCrVthRGJPxAitBeHc6twZWe4tK7Sodvh074K674Ikn0jQ5gCUHY9JAufdBeMngPVXd4T0sOaSA\np5+GevXczJopb/HisCMwxlRRPDfKzRWRNoFHYuIyeTI8/ri7WzqleyyVzQ199dWuyGOMSTsxE4SI\nlFU/HQfMEJFFIjJbROaIyG7zNZjgrV0LHTu6HqHNmoUdTTnKRl5t3txNgm3dV41JS+W1QcwA2gAX\nJykWU46SEpccbroJzj8/7GhisJFXjcko5SUIAVDVJUmKxZTjL39xXVl79w47knJs2uQmvba5oY3J\nCDG7uYozsJ2OAAAVt0lEQVTIKuCpWG9U1ZjrgpDN3VzfeMP1Wpo1C/bZJ+xojDHpJKhurjWAXGx0\n1VBNmQK33ALvvmvJwRiTXOWVIGarasr0XsrGEsS8efDb37oam3POCTuaCIWFbqLrP/0pxbtSGWOq\nU4Ior5ur/eeHaPVquOACGDAgxZJDWQ+loiLrvmpMhiuviqlD0qIwu/j5Z7jsMtch6Oqrw47GYz2U\njMk6MUsQqro+mYEYR9VN+tOiBdx3X9jReL744pf7GubMseRgTJaoymiuJkAvv+yux9OmpVD1/mGH\nuVby448POxJjTBLZlKMpZOlSOPFEmDABjjoq7GiMMZkgqEZqk0Tbt7vhu//855CTQ4YnYWNM/CxB\npIgePWDvveHuu0MMYswYaNvWZStjTNazNogUMGQIfPIJfPYZ5ISRsqPnhk6r6emMMUGxEkTIVq92\n95uNGwf164cQQOTIq9ZDyRgTwRqpQ3b55a7N4aGHQjj43LnQubPrOmWJwZiMVJ1GaksQIXr5ZejX\nD2bPDnHw05IS2MNqGo3JVCndi0lEzhWRhSLytYj09FnfQ0Tmi8hcEflYRA4KOqZUMG8e9OzpqpZC\nHRnbkoMxJoZAE4SI5AADgHOAI4FOInJ41GazgeNV9VjgVeCJIGNKBcXFbjbO/v3h179O0kG//DJJ\nBzLGZIqgSxBtgcWqukJVtwOjgEsiN1DViaq6xXs5HTgg4JhCpQo33wxnnAHXXZeEA5bNDd2pE2zd\nmoQDGmMyRdAJ4gBgZcTrVZSfAG4G3g80opA9+ywsWQL/+EfAB1LdtYfSzJlQu3bABzXGZJKgK6D9\nGkZ8W5pF5DrgeKB9rJ316dNn5/P8/Hzy8/OrF12SffYZPPwwTJ8ecLvD//4Ht98O8+fDm2+68TuM\nMVmhoKCAgoKChOwr0F5MItIO6KOq53qv7wNUVR+L2u4s4B/A6apaFGNfad2LqagI2rRxJYdLLw34\nYGvWwMCB0KuXzQ1tTJZL2W6uIlIDWISbW2INMAPopKoLIrY5DhgLnKOqS8rZV9omiNJSuPBC1yDd\nr1/Y0RhjsknKdnNV1R3AncBHwHxglKouEJG+InKht9njQB1grIjMEZE3gowpDE895Xou/f3vYUdi\njDHxsxvlArZxI7Rs6dodWrZM8M4LC+G556B375AGcTLGpLqULUEYd6/DRRclODmowujRrofSli3u\nbmhjjEkwu402QJMmwaBBrvSQMIWFcMcd1kPJGBM4K0EE5Jtv3P1pI0a42xASYuFCV2po0cKNvGrJ\nwRgTIGuDCMD338Npp8G998KttyZwxzt2uCEzWrdO4E6NMZksZbu5JlK6JIiNG6F9e7jySvjLX8KO\nxhiT7SxBpJDOnd3EPwMHglTpT+IpLbWeScaYarNeTCnitdfccBpPPlmN5BDZQ+nnnxManzHGVIb1\nYkqQF190VUpvvgl77VXFnUT2UBoyBH71q0SGaIwxlWIliAR4/HF47DHXrbVKHYsiSw3WQ8kYkyKs\nDaKa+vd37Q0FBXBAVWeyWLTItWq/9JIlBmNMQlkjdUgmToSOHWHGDDiouhOlWqO0MSYA1kgdgqIi\n6NIFXn45AckBLDkYY1KOXZWqoGza0KuugvPOq+QbZ80KLC5jjEkk68VUBf37w6pVbkbPuJX1UFqw\nwNVJ1akTWHzGGJMIVoKopDFjXIJ49VWoVSuON0T3UJo1y5KDMSYtWCN1JXzzDZx0EnzySZzDIRUV\nucGYyu5rsB5Kxpgkq04jtVUxxWn7drjuOnczXNxj5eXkuJLDiBE2N7QxJu1YCSJOPXu6gVTffts6\nHBlj0oeVIAL2wQcwcqS7wdmSgwnDIYccwooVK8IOw6Swpk2bsnz58oTu00oQFZg5Ey68EMaOhdNP\nj7FRYSH06wcPPxxny7UxleN9Cww7DJPCYn1G7Ea5gHz1lUsOgwfHSA6RPZRE3GtjjMkQVsUUw9at\ncPXV8Pe/uySxG5sb2hiT4awEEcPf/gYtW8KNN/qsXL7cRl41xmQ8a4Pw8e23cOyxrtdSkyY+G6i6\nO6J//eukxGOMtUGYigTRBmEJwkeHDnD22a5rqzGpwBKEqUhaNlKLyLkislBEvhaR3S65IlJLREaJ\nyGIRmSYiBwcdU3nWrnWjYfTo4S0oKQkzHGPSwqhRo2jXrh25ubk0atSIk046iUGDBu1cv3r1aq68\n8kr23Xdf8vLyaN26NcOGDavUMYYMGUJOTg7jxo3bZfnQoUM57bTTdtu+WbNm/Oc//9n5esaMGVxw\nwQXk5eWxzz770K5dO4YMGVK5E41D//79ady4MXl5efzud79j+/btvtuNHDmSunXrUq9ePerVq0ed\nOnXIyclhzpw5APTr14+jjz6aevXq0aJFC/r165fwWCsSaIIQkRxgAHAOcCTQSUQOj9rsZmC9qh4K\nPA08HmRMFRkyBM4/H2rV9HooHX44FBeHGZIxKe3JJ5+kR48e9OzZk7Vr1/L999/z/PPPM3Xq1J0X\nxy5dutC0aVNWrlxJUVERw4YNY//996/UcYYNG0bDhg0ZOnTobuukgkngp02bRocOHTjjjDNYsmQJ\n69atY9CgQXzwwQeViqEiH374IY8//jgTJkxg+fLlLFmyhN69e/tu27lzZzZt2kRxcTHFxcUMHDiQ\nFi1acNxxx+3cZvjw4WzcuJH333+fAQMGMKZSI4QmgKoG9gDaAe9HvL4P6Bm1zQfAid7zGsD/YuxL\ng7Zjh2penuo3U9eqXnGF6uGHq06fHvhxjalIMj7/VfHDDz9onTp19PXXXy93u9zcXJ03b16Vj7N8\n+XKtUaOGvvbaa1qzZk0tLCzcuW7IkCF62mmn7faeQw45RMePH6+qqqeeeqp269atysePV+fOnbVX\nr147X48fP14bNWoU13vPOOMMfeihh2Ku7969u3bv3j3m+lifEW95la7hQVcxHQCsjHi9ylvmu42q\n7gA2ikiDgOPytXyZcrWOpsVl1kPJmHhMmzaNbdu2cfHFF5e73UknncQdd9zB6NGjWblyZbnb+hk2\nbBgnnHACl112GYcffjj/+te/4n7vzz//zLRp07jiiivifs+UKVPIy8ujQYMG5OXl7fK8QYMGTJ06\n1fd98+fPp3XEYG2tW7emsLCQDRs2lHu8FStWMGnSJK6//vqY20yaNIkjjzwy7nNIhKAThF+5L7oV\nJXob8dkmKd4euJL7cx519zU89pgNsGfShkhiHpW1bt069tlnH3IixqA55ZRTyMvLY6+99mLy5MkA\njB07ltNPP52HH36Y5s2b06ZNGz7//PO4jzN8+HCuvfZawFXN+FUzxbJhwwZKS0tp3Lhx3O855ZRT\n2LBhA+vXr2fDhg27PF+/fj0nn3yy7/s2b95M/fr1d76uX78+qsqmTZvKPd6wYcM47bTTaNq0qe/6\n3r17o6rc6NvvPjhB3yi3CohsdD4Q+C5qm5XAQcB3IlIDqKeqvum2T58+O5/n5+eTn5+fyFi54cGD\nqdFnNtStUoO/MaEJq4NTw4YNWbduHaWlpTuTxJQpUwA4+OCDKS0tBdyF8pFHHuGRRx5h/fr13HPP\nPVx22WVxlSamTJnCsmXLuOaaawDo1KkTvXr14osvvuCYY45hjz328G0I3r59OzVr1iQvL4+cnBzW\nrFlDq1atEnXqvnJzcymOaLMsLi5GRKhbt2657xs+fDgPPPCA77oBAwYwYsQIJk+eTM2aNSuMoaCg\ngIKCgkrFHVNV66bieeDaFL4BmgK1gLnAEVHb3AEM9J53BEbF2FfMujdjMl2qfv43btyoubm5+tpr\nr+227sADD9SJEyf6vu/LL7/UnJwcXb9+fYXHuOWWW3SPPfbQxo0ba6NGjbRRo0Zao0YNveeee1RV\n9dNPP9UmTZrs8p4ff/xRa9asqd9++62qVr4NYtKkSZqbm6t169bd5VG2bPLkyb7v69y5sz7wwAM7\nX48fP14bN25c7rEmT56subm5unnz5t3WDR48WA866CBdvnx5hTHH+oxQjTaIQBOEi41zgUXAYuA+\nb1lf4ELveW1gjLd+OnBIjP1U+AsyJlOl8uf/8ccf10aNGum4ceN08+bNWlpaqnPmzNEGDRrsTBA9\ne/bUL7/8UktKSrS4uFjvuOMObdWq1c59HHLIITp06NDd9r1lyxbde++99ZVXXtG1a9fufDz33HO6\n//77644dO3Tr1q3avHlzffTRR3XLli26efNm7datm5588sk79zN16lStW7eu9uvXT4uKilRVde7c\nudqxY8eE/i4++OADbdy4sX711Ve6fv16PfPMM/X+++8v9z233HKLdu3adbflI0aM0EaNGunChQvj\nOnZaJohEPVL5H8SYoKX653/kyJHatm1brVOnju63337arl07femll3T79u2qqtqtWzc99NBDtW7d\nurrffvvpRRddtPPCt23bNq1Xr54uWrRot/2OGjVKmzRpoiUlJbss37Jli+6777767rvvqqrqggUL\n9JxzztF99tlHGzVqpFdddZWuWrVql/fMnDlTzzvvPN177721YcOG2q5dOx0+fHjCfxf9+/fX/fff\nX+vXr68333yzbtu2bee6I488UkeOHLnLeeTl5emECRN220+zZs20Vq1au5Rcbr/99pjHDSJB2J3U\nxqSBTL6TesqUKQwcOLBSPZPM7myojTSJ1ZhEy+QEYRIjLYfaMMYYk54sQRhjjPFlCcIYY4wvSxDG\nGGN8WYIwxhjjy+akNiYNNG3atMIhrU12izWOU3VYN1djjMlg1s01gRI2yFUKyKRzgcw6n0w6F8is\n88mkc6kuSxBRMunDkUnnApl1Ppl0LpBZ55NJ51JdliCMMcb4sgRhjDHGV1o1UocdgzHGpKOMH6zP\nGGNMclkVkzHGGF+WIIwxxvjK2gQhIueKyEIR+VpEevqsryUio0RksYhME5GDw4gzHnGcSw8RmS8i\nc0XkYxE5KIw441XR+URsd6WIlIpIm2TGVxnxnIuIXO39ff4rIiOSHWNlxPFZO0hE/iMis73P23lh\nxBkPERksImtF5ItytnnGuwbMFZFjkxlfSqjqVHTp/MAlxm+ApkBNYC5weNQ2twMDvefXAKPCjrsa\n59Ie2NN7fluqnku85+NtlwtMBKYCbcKOuxp/m5bALKCe93qfsOOu5vm8ANzqPT8CWBZ23OWcz6nA\nscAXMdafB7zrPT8RmB52zMl+ZGsJoi2wWFVXqOp2YBRwSdQ2lwBDvefjgA5JjK8yKjwXVZ2oqlu8\nl9OBA5IcY2XE87cB+CvwGLA1mcFVUjzncgvwnKoWA6jquiTHWBnxnE8pUM97vjewOonxVYqqTgY2\nlLPJJcAwb9vPgPoisn8yYksV2ZogDgBWRrxexe4XzZ3bqOoOYKOINEhOeJUSz7lEuhl4P9CIqqfC\n8/GK+geq6nvJDKwK4vnbtAIOE5HJIjJVRM5JWnSVF8/59AW6iMhK4B2gW5JiC0L0+a4mtb9cJVy2\njubq1yc4ur9v9Dbis00qiOdc3IYi1wHH46qcUlW55yNuSNP+QNcK3pMK4vnb7IGrZjodOBiYJCJH\nlpUoUkw859MJeEVV+4tIO2AEcGTgkQUj7v+tTJWtJYhVuH/GMgcC30VtsxI4CEBEauDqiMsrjoYl\nnnNBRM4C/gxc5FUPpKqKzqcu7oJTICLLgHbAmynaUB3P32YV8KaqlqrqcmARcGhywqu0eM7nZmAM\ngKpOB/YUkX2SE17CrcK7Bnh8/7cyWbYmiJlASxFpKiK1gI7AW1HbvM0v31KvAv6TxPgqo8JzEZHj\ngOeBi1W1KIQYK6Pc81HVYlXdT1Wbq2ozXJvKRao6O6R4yxPP5+wN4EwA70J6KLA0qVHGL57zWQGc\nBSAiRwC1U7xdRYhdAn0LuB7AKw1tVNW1yQosFWRlFZOq7hCRO4GPcElysKouEJG+wExVfQcYDAwX\nkcVAEe6fIeXEeS6PA3WAsV4VzQpVvTS8qGOL83x2eQspWsUUz7mo6ocicraIzAdKgHtTtKQa79/m\nXuBFEemBa7DuGnuP4RKRkUA+0FBEvgV6A7UAVdV/qup7InK+iHwD/AjcGF604bChNowxxvjK1iom\nY4wxFbAEYYwxxpclCGOMMb4sQRhjjPFlCcIYY4wvSxDGGGN8WYIwKUNEdnjDRM/xfsYcYt27Weu/\nCTjmBG/46rkiMklEKn0Xs4jc6g1jgoh0FZFGEev+KSKHJzjOz0TkmDjec5eI7FndY5vsZQnCpJIf\nVbWNqh7n/fy2gu0TdRNPJ1U9FjdyZ7/KvllVX1DVsnkcbiBiQDdV/b2qLkxIlL/EOYj44rwb2CtB\nxzZZyBKESSW73RHtlRQ+FZHPvUc7n21+7X2rLpukpoW3/NqI5YO8u8jLO+6nQNl7O3jvmyciL4lI\nTW/5o/LL5EuPe8t6i8g9InIFcAIwwnvvnt43/zYicpuIPBYRc1cR+UcV45wGNInY10ARmSFuwqHe\n3rJu3jYTRGS8t+xsb8TYz0VktIhY8jDlsgRhUsmvIqqYXvWWrQXOUtUTcMOdPOvzvtuAp1W1De4C\nvcqr1rkGONlbXgpcW8HxLwb+KyK1gVeAq1S1NW5ynNtFJA+4VFWP9L7JPxzxXlXVV4HPgc5eCWhL\nxPpxwOURr68BRlcxznNxYziVuV9V2wKtgXwROUpVn8UNT52vqh1EpCHQC+jg/S5nAfdUcByT5bJy\nLCaTsn7yLpKRagEDxM0BsQP/kU6nAb3ETaX6mqp+IyIdgDbATO8b+Z64ZOPnXyLyM7AcN3/BYcBS\nVV3irR8K3AE8B/wsIi8C7+HmO/CzWwlAVdeJyBIRaYubla2Vqk4VkT9UMs7auHG1Iqe/7Cgit+D+\nnxsBvwa+ZNeB6Np5y6d4x6mJ+70ZE5MlCJPqegDfq+ox4oZd/zl6A1X9t4hMBy4E3hWRW3EXxqGq\n2iuOY3RW1TllL8SNqup3kd/hXeA74Eb4vZPKzTQ4BldaWAi8Xna4ysbpVW0NAK4QkUNwJYHjVbVY\nRF7BJZloAnykqhWVTozZyaqYTCrxq3uvD6zxnl8P1NjtTSLNVHWZV63yFnAMMB64UkT29bbJK6dX\nVPRxFwJNRaS597oLMNGrs99bVT8A/ugdJ9omfplyM9prwKW4qrLR3rKqxPkgcKKIHOYdazOwSdx0\nmOdFbF8cEct04JSI9plfVaXHlskuliBMKvHrlTQQuEFE5uCm5/zRZ5trRORLb5sjgWGqugB4APhI\nRObhhqhu5PPe3Y6pqltxQzuP8967AzefRj3gHW/Zp7jSTbQhwPNljdSR+1fVjcBXwMGq+rm3rNJx\nem0bT+KGBv8CmAsswM3eNjniPS8C74vIeG9OhhuBf3vHmYarSjMmJhvu2xhjjC8rQRhjjPFlCcIY\nY4wvSxDGGGN8WYIwxhjjyxKEMcYYX5YgjDHG+LIEYYwxxpclCGOMMb7+H/YxdhwQi4EpAAAAAElF\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5d2ee2b8d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Smart link 0.8\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, 'b',\n",
    "label='GS, AUC = %0.2f'% roc_auc)\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.1,1.2])\n",
    "plt.ylim([-0.1,1.2])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMcbX/wNVn1tpxjSX8wAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5d2ee967d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Smart link 0.5\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, 'b',\n",
    "label='GS, AUC = %0.2f'% roc_auc)\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.1,1.2])\n",
    "plt.ylim([-0.1,1.2])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}