Edge_attack.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\n",
    "from Lopacity import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Naive anonymization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 104 ms, sys: 0 ns, total: 104 ms\n",
      "Wall time: 100 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Naive data\n",
    "work_dir = \"data/USpowerGrid/\"\n",
    "file_name = work_dir + 'USpowerGrid.mtx'\n",
    "G = read_mtx_graph(file_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from attack import *\n",
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "alpha = 0.25"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "G1, G2 = edge_split(G, alpha)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.0987903225806\n",
      "Finished : 0.197580645161\n",
      "Finished : 0.296370967742\n",
      "Finished : 0.395161290323\n",
      "Finished : 0.493951612903\n",
      "Finished : 0.592741935484\n",
      "Finished : 0.691532258065\n",
      "Finished : 0.790322580645\n",
      "Finished : 0.889112903226\n",
      "Finished : 0.987903225806\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0969696969697\n",
      "Finished : 0.193939393939\n",
      "Finished : 0.290909090909\n",
      "Finished : 0.387878787879\n",
      "Finished : 0.484848484848\n",
      "Finished : 0.581818181818\n",
      "Finished : 0.678787878788\n",
      "Finished : 0.775757575758\n",
      "Finished : 0.872727272727\n",
      "Finished : 0.969696969697\n",
      "Aux finished 0.162354946136 sec\n",
      "Start adding positive examples\n",
      "Finished : 0.0989660265879\n",
      "Finished : 0.197932053176\n",
      "Finished : 0.296898079764\n",
      "Finished : 0.395864106352\n",
      "Finished : 0.494830132939\n",
      "Finished : 0.593796159527\n",
      "Finished : 0.692762186115\n",
      "Finished : 0.791728212703\n",
      "Finished : 0.890694239291\n",
      "Finished : 0.989660265879\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.1\n",
      "Finished : 0.2\n",
      "Finished : 0.3\n",
      "Finished : 0.4\n",
      "Finished : 0.5\n",
      "Finished : 0.6\n",
      "Finished : 0.7\n",
      "Finished : 0.8\n",
      "Finished : 0.9\n",
      "San finished 0.178848028183 sec\n",
      "CPU times: user 820 ms, sys: 12 ms, total: 832 ms\n",
      "Wall time: 819 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "X_naive, Y_naive = gen_trainset(G1, G2, alpha, work_dir + \"train_naive.dump\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.0998162890386\n",
      "Finished : 0.199632578077\n",
      "Finished : 0.299448867116\n",
      "Finished : 0.399265156154\n",
      "Finished : 0.499081445193\n",
      "Finished : 0.598897734231\n",
      "Finished : 0.69871402327\n",
      "Finished : 0.798530312309\n",
      "Finished : 0.898346601347\n",
      "Finished : 0.998162890386\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0988764044944\n",
      "Finished : 0.197752808989\n",
      "Finished : 0.296629213483\n",
      "Finished : 0.395505617978\n",
      "Finished : 0.494382022472\n",
      "Finished : 0.593258426966\n",
      "Finished : 0.692134831461\n",
      "Finished : 0.791011235955\n",
      "Finished : 0.889887640449\n",
      "Finished : 0.988764044944\n",
      "Test set generated in  0.509311914444 sec\n"
     ]
    }
   ],
   "source": [
    "X_t_naive, Y_t_naive = gen_testset(G1, G2, work_dir + \"test_naive.dump\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Deanonymization score\n",
      "0.822425409047\n",
      "CPU times: user 2.22 s, sys: 96 ms, total: 2.32 s\n",
      "Wall time: 1.85 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest.fit(X_naive, Y_naive)\n",
    "print(\"Deanonymization score\")\n",
    "print(forest.score(X_t_naive, Y_t_naive))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Link Anonymized data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 24 ms, sys: 12 ms, total: 36 ms\n",
      "Wall time: 30 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Naive data\n",
    "anon_data = [\"USpowerGrid.mtx_2_0.900000.txt\", \"USpowerGrid.mtx_2_0.800000.txt\", \"USpowerGrid.mtx_3_0.700000.txt\"]\n",
    "file_name_anon = work_dir + anon_data[0]\n",
    "G_anon = read_mtx_graph(file_name_anon)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6594"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "G.number_of_edges()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6594"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "G_anon.number_of_edges()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from utils import calc_cc, calc_dist, calc_emd, number_of_fake_edges"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0\n"
     ]
    }
   ],
   "source": [
    "print(calc_cc(G, G_anon))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.0\n"
     ]
    }
   ],
   "source": [
    "print(calc_dist(G, G_anon))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0\n"
     ]
    }
   ],
   "source": [
    "print(number_of_fake_edges(G, G_anon))"
   ]
  },
  {
   "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "G1_a, G2_a = edge_split(G_anon, alpha)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.0997920997921\n",
      "Finished : 0.199584199584\n",
      "Finished : 0.299376299376\n",
      "Finished : 0.399168399168\n",
      "Finished : 0.49896049896\n",
      "Finished : 0.598752598753\n",
      "Finished : 0.698544698545\n",
      "Finished : 0.798336798337\n",
      "Finished : 0.898128898129\n",
      "Finished : 0.997920997921\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0945945945946\n",
      "Finished : 0.189189189189\n",
      "Finished : 0.283783783784\n",
      "Finished : 0.378378378378\n",
      "Finished : 0.472972972973\n",
      "Finished : 0.567567567568\n",
      "Finished : 0.662162162162\n",
      "Finished : 0.756756756757\n",
      "Finished : 0.851351351351\n",
      "Finished : 0.945945945946\n",
      "Aux finished 0.163866043091 sec\n",
      "Start adding positive examples\n",
      "Finished : 0.0990566037736\n",
      "Finished : 0.198113207547\n",
      "Finished : 0.297169811321\n",
      "Finished : 0.396226415094\n",
      "Finished : 0.495283018868\n",
      "Finished : 0.594339622642\n",
      "Finished : 0.693396226415\n",
      "Finished : 0.792452830189\n",
      "Finished : 0.891509433962\n",
      "Finished : 0.990566037736\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0966386554622\n",
      "Finished : 0.193277310924\n",
      "Finished : 0.289915966387\n",
      "Finished : 0.386554621849\n",
      "Finished : 0.483193277311\n",
      "Finished : 0.579831932773\n",
      "Finished : 0.676470588235\n",
      "Finished : 0.773109243697\n",
      "Finished : 0.86974789916\n",
      "Finished : 0.966386554622\n",
      "San finished 0.170298099518 sec\n",
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.0998104864182\n",
      "Finished : 0.199620972836\n",
      "Finished : 0.299431459255\n",
      "Finished : 0.399241945673\n",
      "Finished : 0.499052432091\n",
      "Finished : 0.598862918509\n",
      "Finished : 0.698673404927\n",
      "Finished : 0.798483891346\n",
      "Finished : 0.898294377764\n",
      "Finished : 0.998104864182\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.0989010989011\n",
      "Finished : 0.197802197802\n",
      "Finished : 0.296703296703\n",
      "Finished : 0.395604395604\n",
      "Finished : 0.494505494505\n",
      "Finished : 0.593406593407\n",
      "Finished : 0.692307692308\n",
      "Finished : 0.791208791209\n",
      "Finished : 0.89010989011\n",
      "Finished : 0.989010989011\n",
      "Test set generated in  0.531386137009 sec\n",
      "CPU times: user 1.88 s, sys: 36 ms, total: 1.92 s\n",
      "Wall time: 1.9 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "X_an, Y_an = gen_trainset(G1_a, G2_a, alpha, work_dir + \"train_anon.dump\")\n",
    "X_t_an, Y_t_an = gen_testset(G1_a, G2_a, work_dir + \"test_naive.dump\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Deanonymization score\n",
      "0.817303044949\n",
      "CPU times: user 2 s, sys: 92 ms, total: 2.09 s\n",
      "Wall time: 1.58 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "forest_a = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest_a.fit(X_an, Y_an)\n",
    "print(\"Deanonymization score\")\n",
    "print(forest.score(X_t_an, Y_t_an))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Deanonymization score\n",
      "0.830225711482\n",
      "CPU times: user 2 s, sys: 140 ms, total: 2.14 s\n",
      "Wall time: 1.65 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "forest_a = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest_a.fit(X_an, Y_an)\n",
    "print(\"Deanonymization score\")\n",
    "print(forest.score(X_t_an, Y_t_an))"
   ]
  },
  {
   "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def anonimize(g):\n",
    "    t1 = time()\n",
    "    result_optval2 = intersect_anonimize(g, L, theta, score='opt2', mode = 'val')\n",
    "    tot_time = time()-t1\n",
    "    print(\"Total time \"+ str(tot_time))\n",
    "\n",
    "    g1 = result_optval2['new_graph']\n",
    "    dist = (g.number_of_edges() - g1.number_of_edges())/g.number_of_edges()\n",
    "    print(\"Distortion \"+ str(dist))\n",
    "    return g1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def anonimize_orig(g):\n",
    "    t1 = time()\n",
    "    result_lomax = anonimize_lopacity(g=g, L=L, theta=theta, mode='max')\n",
    "    print(\"Total time \"+ str(time()-t1))\n",
    "\n",
    "    # Storing results:\n",
    "    g1 = result_lomax['new_graph']\n",
    "    dist = (g.number_of_edges() - g1.number_of_edges())/g.number_of_edges()\n",
    "    print(\"Distortion \"+ str(dist))\n",
    "    return g1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Edges removed: \n",
      "[(6, 144)]\n",
      "Edges removed: \n",
      "[(13, 123)]\n",
      "Edges removed: \n",
      "[(13, 87)]\n",
      "Edges removed: \n",
      "[(5, 104)]\n",
      "Edges removed: \n",
      "[(15, 198)]\n",
      "Edges removed: \n",
      "[(0, 198)]\n",
      "Edges removed: \n",
      "[(2, 135)]\n",
      "Edges removed: \n",
      "[(4, 135)]\n",
      "Edges removed: \n",
      "[(0, 156)]\n",
      "Edges removed: \n",
      "[(2, 144)]\n",
      "Edges removed: \n",
      "[(2, 155)]\n",
      "Edges removed: \n",
      "[(2, 198)]\n",
      "Edges removed: \n",
      "[(3, 173)]\n",
      "Edges removed: \n",
      "[(24, 29)]\n",
      "Edges removed: \n",
      "[(19, 106)]\n",
      "Edges removed: \n",
      "[(12, 135)]\n",
      "Edges removed: \n",
      "[(2, 156)]\n",
      "Edges removed: \n",
      "[(3, 156)]\n",
      "Edges removed: \n",
      "[(20, 156)]\n",
      "Edges removed: \n",
      "[(19, 156)]\n",
      "Edges removed: \n",
      "[(2, 191)]\n",
      "Edges removed: \n",
      "[(5, 191)]\n",
      "Edges removed: \n",
      "[(47, 198)]\n",
      "Edges removed: \n",
      "[(15, 170)]\n",
      "Edges removed: \n",
      "[(24, 156)]\n",
      "Edges removed: \n",
      "[(8, 135)]\n",
      "Edges removed: \n",
      "[(2, 48)]\n",
      "Edges removed: \n",
      "[(24, 3)]\n",
      "Edges removed: \n",
      "[(15, 191)]\n",
      "Edges removed: \n",
      "[(5, 69)]\n",
      "Edges removed: \n",
      "[(0, 141)]\n",
      "Edges removed: \n",
      "[(18, 141)]\n",
      "Edges removed: \n",
      "[(12, 141)]\n",
      "Edges removed: \n",
      "[(9, 141)]\n",
      "Edges removed: \n",
      "[(4, 32)]\n",
      "Edges removed: \n",
      "[(7, 84)]\n",
      "Edges removed: \n",
      "[(17, 144)]\n",
      "Edges removed: \n",
      "[(7, 113)]\n",
      "Edges removed: \n",
      "[(20, 45)]\n",
      "Edges removed: \n",
      "[(0, 89)]\n",
      "Edges removed: \n",
      "[(30, 11)]\n",
      "Edges removed: \n",
      "[(21, 144)]\n",
      "Edges removed: \n",
      "[(5, 55)]\n",
      "Edges removed: \n",
      "[(7, 78)]\n",
      "Edges removed: \n",
      "[(28, 107)]\n",
      "Edges removed: \n",
      "[(5, 11)]\n",
      "Edges removed: \n",
      "[(0, 155)]\n",
      "Edges removed: \n",
      "[(9, 41)]\n",
      "Edges removed: \n",
      "[(7, 21)]\n",
      "Edges removed: \n",
      "[(5, 78)]\n",
      "Edges removed: \n",
      "[(39, 87)]\n",
      "Edges removed: \n",
      "[(25, 89)]\n",
      "Edges removed: \n",
      "[(20, 151)]\n",
      "Edges removed: \n",
      "[(28, 101)]\n",
      "Edges removed: \n",
      "[(17, 69)]\n",
      "Edges removed: \n",
      "[(1, 108)]\n",
      "Edges removed: \n",
      "[(12, 60)]\n",
      "Edges removed: \n",
      "[(21, 106)]\n",
      "Edges removed: \n",
      "[(4, 106)]\n",
      "Edges removed: \n",
      "[(3, 106)]\n",
      "Edges removed: \n",
      "[(21, 46)]\n",
      "Edges removed: \n",
      "[(4, 46)]\n",
      "Edges removed: \n",
      "[(26, 106)]\n",
      "Edges removed: \n",
      "[(28, 52)]\n",
      "Edges removed: \n",
      "[(0, 52)]\n",
      "Edges removed: \n",
      "[(7, 52)]\n",
      "Edges removed: \n",
      "[(14, 106)]\n",
      "Edges removed: \n",
      "[(0, 31)]\n",
      "Edges removed: \n",
      "[(4, 51)]\n",
      "Edges removed: \n",
      "[(20, 35)]\n",
      "Edges removed: \n",
      "[(2, 34)]\n",
      "Edges removed: \n",
      "[(1, 47)]\n",
      "Edges removed: \n",
      "[(0, 59)]\n",
      "Edges removed: \n",
      "[(25, 181)]\n",
      "Edges removed: \n",
      "[(1, 9)]\n",
      "Edges removed: \n",
      "[(20, 28)]\n",
      "Edges removed: \n",
      "[(17, 2)]\n",
      "Edges removed: \n",
      "[(28, 181)]\n",
      "Edges removed: \n",
      "[(32, 57)]\n",
      "Edges removed: \n",
      "[(12, 124)]\n",
      "Edges removed: \n",
      "[(55, 106)]\n",
      "Edges removed: \n",
      "[(16, 27)]\n",
      "Edges removed: \n",
      "[(3, 40)]\n",
      "Edges removed: \n",
      "[(3, 12)]\n",
      "Edges removed: \n",
      "[(16, 42)]\n",
      "Edges removed: \n",
      "[(10, 97)]\n",
      "Edges removed: \n",
      "[(13, 47)]\n",
      "Edges removed: \n",
      "[(13, 39)]\n",
      "Edges removed: \n",
      "[(18, 168)]\n",
      "Edges removed: \n",
      "[(4, 27)]\n",
      "Edges removed: \n",
      "[(28, 168)]\n",
      "Edges removed: \n",
      "[(1, 59)]\n",
      "Edges removed: \n",
      "[(6, 76)]\n",
      "Edges removed: \n",
      "[(12, 14)]\n",
      "Edges removed: \n",
      "[(1, 94)]\n",
      "Edges removed: \n",
      "[(12, 94)]\n",
      "Edges removed: \n",
      "[(4, 108)]\n",
      "Edges removed: \n",
      "[(0, 41)]\n",
      "Edges removed: \n",
      "[(19, 108)]\n",
      "Edges removed: \n",
      "[(28, 21)]\n",
      "Edges removed: \n",
      "[(15, 21)]\n",
      "Edges removed: \n",
      "[(15, 80)]\n",
      "Edges removed: \n",
      "[(7, 104)]\n",
      "Edges removed: \n",
      "[(25, 94)]\n",
      "Edges removed: \n",
      "[(26, 38)]\n",
      "Edges removed: \n",
      "[(15, 94)]\n",
      "Edges removed: \n",
      "[(9, 25)]\n",
      "Edges removed: \n",
      "[(21, 108)]\n",
      "Edges removed: \n",
      "[(6, 104)]\n",
      "Edges removed: \n",
      "[(31, 108)]\n",
      "Edges removed: \n",
      "[(29, 38)]\n",
      "Edges removed: \n",
      "[(12, 18)]\n",
      "Edges removed: \n",
      "[(28, 29)]\n",
      "Edges removed: \n",
      "[(7, 8)]\n",
      "Edges removed: \n",
      "[(50, 48)]\n",
      "Edges removed: \n",
      "[(28, 64)]\n",
      "Edges removed: \n",
      "[(15, 0)]\n",
      "Edges removed: \n",
      "[(29, 80)]\n",
      "Edges removed: \n",
      "[(25, 50)]\n",
      "Edges removed: \n",
      "[(20, 104)]\n",
      "Edges removed: \n",
      "[(19, 111)]\n",
      "Edges removed: \n",
      "[(10, 41)]\n",
      "Edges removed: \n",
      "[(10, 15)]\n",
      "Edges removed: \n",
      "[(1, 32)]\n",
      "Edges removed: \n",
      "[(28, 66)]\n",
      "Edges removed: \n",
      "[(12, 115)]\n",
      "Edges removed: \n",
      "[(17, 36)]\n",
      "Edges removed: \n",
      "[(41, 51)]\n",
      "Edges removed: \n",
      "[(7, 105)]\n",
      "Edges removed: \n",
      "[(18, 51)]\n",
      "Edges removed: \n",
      "[(17, 184)]\n",
      "Edges removed: \n",
      "[(3, 18)]\n",
      "Edges removed: \n",
      "[(0, 74)]\n",
      "Edges removed: \n",
      "[(6, 74)]\n",
      "Edges removed: \n",
      "[(18, 184)]\n",
      "Edges removed: \n",
      "[(6, 19)]\n",
      "Edges removed: \n",
      "[(28, 43)]\n",
      "Edges removed: \n",
      "[(6, 88)]\n",
      "Edges removed: \n",
      "[(2, 3)]\n",
      "Edges removed: \n",
      "[(29, 152)]\n",
      "Edges removed: \n",
      "[(28, 88)]\n",
      "Edges removed: \n",
      "[(68, 43)]\n",
      "Edges removed: \n",
      "[(72, 34)]\n",
      "Edges removed: \n",
      "[(14, 147)]\n",
      "Edges removed: \n",
      "[(27, 108)]\n",
      "Edges removed: \n",
      "[(72, 107)]\n",
      "Edges removed: \n",
      "[(35, 34)]\n",
      "Edges removed: \n",
      "[(32, 33)]\n",
      "Edges removed: \n",
      "[(120, 31)]\n",
      "Edges removed: \n",
      "[(7, 34)]\n",
      "Edges removed: \n",
      "[(115, 40)]\n",
      "Edges removed: \n",
      "[(13, 95)]\n",
      "Edges removed: \n",
      "[(21, 37)]\n",
      "Edges removed: \n",
      "[(21, 195)]\n",
      "Edges removed: \n",
      "[(28, 59)]\n",
      "Edges removed: \n",
      "[(4, 47)]\n",
      "Edges removed: \n",
      "[(4, 155)]\n",
      "Edges removed: \n",
      "[(9, 123)]\n",
      "Edges removed: \n",
      "[(2, 51)]\n",
      "Edges removed: \n",
      "[(29, 62)]\n",
      "Edges removed: \n",
      "[(29, 108)]\n",
      "Edges removed: \n",
      "[(9, 15)]\n",
      "Edges removed: \n",
      "[(12, 28)]\n",
      "Edges removed: \n",
      "[(28, 37)]\n",
      "Edges removed: \n",
      "[(32, 74)]\n",
      "Edges removed: \n",
      "[(3, 37)]\n",
      "Edges removed: \n",
      "[(1, 33)]\n",
      "Edges removed: \n",
      "[(24, 59)]\n",
      "Edges removed: \n",
      "[(2, 108)]\n",
      "Edges removed: \n",
      "[(3, 181)]\n",
      "Edges removed: \n",
      "[(8, 59)]\n",
      "Edges removed: \n",
      "[(8, 95)]\n",
      "Edges removed: \n",
      "[(2, 4)]\n",
      "Edges removed: \n",
      "[(13, 166)]\n",
      "Edges removed: \n",
      "[(17, 118)]\n",
      "Edges removed: \n",
      "[(8, 172)]\n",
      "Edges removed: \n",
      "[(162, 45)]\n",
      "Edges removed: \n",
      "[(36, 74)]\n",
      "Edges removed: \n",
      "[(12, 137)]\n",
      "Edges removed: \n",
      "[(5, 109)]\n",
      "Edges removed: \n",
      "[(3, 77)]\n",
      "Edges removed: \n",
      "[(24, 57)]\n",
      "Edges removed: \n",
      "[(1, 190)]\n",
      "Edges removed: \n",
      "[(194, 62)]\n",
      "Edges removed: \n",
      "[(14, 121)]\n",
      "Edges removed: \n",
      "[(5, 86)]\n",
      "Edges removed: \n",
      "[(13, 195)]\n",
      "Edges removed: \n",
      "[(121, 24)]\n",
      "Edges removed: \n",
      "[(15, 20)]\n",
      "Edges removed: \n",
      "[(10, 81)]\n",
      "Edges removed: \n",
      "[(119, 5)]\n",
      "Edges removed: \n",
      "[(7, 73)]\n",
      "Edges removed: \n",
      "[(80, 31)]\n",
      "Edges removed: \n",
      "[(12, 100)]\n",
      "Edges removed: \n",
      "[(2, 66)]\n",
      "Edges removed: \n",
      "[(185, 31)]\n",
      "Edges removed: \n",
      "[(13, 31)]\n",
      "Edges removed: \n",
      "[(19, 66)]\n",
      "Edges removed: \n",
      "[(5, 56)]\n",
      "Edges removed: \n",
      "[(8, 144)]\n",
      "Edges removed: \n",
      "[(5, 168)]\n",
      "Edges removed: \n",
      "[(5, 33)]\n",
      "Edges removed: \n",
      "[(9, 78)]\n",
      "Edges removed: \n",
      "[(9, 12)]\n",
      "Edges removed: \n",
      "[(20, 22)]\n",
      "Edges removed: \n",
      "[(20, 6)]\n",
      "Edges removed: \n",
      "[(20, 17)]\n",
      "Edges removed: \n",
      "[(6, 62)]\n",
      "Edges removed: \n",
      "[(2, 89)]\n",
      "Edges removed: \n",
      "[(6, 43)]\n",
      "Edges removed: \n",
      "[(0, 50)]\n",
      "Edges removed: \n",
      "[(0, 80)]\n",
      "Edges removed: \n",
      "[(5, 50)]\n",
      "Edges removed: \n",
      "[(80, 25)]\n",
      "Edges removed: \n",
      "[(120, 9)]\n",
      "Edges removed: \n",
      "[(178, 43)]\n",
      "Edges removed: \n",
      "[(26, 25)]\n",
      "Edges removed: \n",
      "[(25, 117)]\n",
      "Edges removed: \n",
      "[(159, 14)]\n",
      "Edges removed: \n",
      "[(25, 56)]\n",
      "Edges removed: \n",
      "[(41, 74)]\n",
      "Edges removed: \n",
      "[(1, 134)]\n",
      "Edges removed: \n",
      "[(1, 119)]\n",
      "Edges removed: \n",
      "[(111, 8)]\n",
      "Edges removed: \n",
      "[(17, 127)]\n",
      "Edges removed: \n",
      "[(19, 57)]\n",
      "Edges removed: \n",
      "[(123, 11)]\n",
      "Edges removed: \n",
      "[(5, 53)]\n",
      "Edges removed: \n",
      "[(8, 18)]\n",
      "Edges removed: \n",
      "[(140, 42)]\n",
      "Edges removed: \n",
      "[(26, 59)]\n",
      "Edges removed: \n",
      "[(43, 17)]\n",
      "Edges removed: \n",
      "[(41, 118)]\n",
      "Edges removed: \n",
      "[(0, 18)]\n",
      "Edges removed: \n",
      "[(79, 49)]\n",
      "Edges removed: \n",
      "[(187, 27)]\n",
      "Edges removed: \n",
      "[(85, 17)]\n",
      "Edges removed: \n",
      "[(67, 29)]\n",
      "Edges removed: \n",
      "[(8, 164)]\n",
      "Edges removed: \n",
      "[(116, 3)]\n",
      "Edges removed: \n",
      "[(21, 115)]\n",
      "Edges removed: \n",
      "[(0, 154)]\n",
      "Edges removed: \n",
      "[(125, 59)]\n",
      "Edges removed: \n",
      "[(17, 26)]\n",
      "Edges removed: \n",
      "[(159, 29)]\n",
      "Edges removed: \n",
      "[(6, 124)]\n",
      "Edges removed: \n",
      "[(79, 2)]\n",
      "Edges removed: \n",
      "[(5, 37)]\n",
      "Edges removed: \n",
      "[(184, 22)]\n",
      "Edges removed: \n",
      "[(14, 92)]\n",
      "Edges removed: \n",
      "[(15, 128)]\n",
      "Edges removed: \n",
      "[(13, 177)]\n",
      "Edges removed: \n",
      "[(163, 4)]\n",
      "Edges removed: \n",
      "[(82, 29)]\n",
      "Edges removed: \n",
      "[(17, 95)]\n",
      "Edges removed: \n",
      "[(42, 17)]\n",
      "Edges removed: \n",
      "[(12, 32)]\n",
      "Edges removed: \n",
      "[(76, 32)]\n",
      "Edges removed: \n",
      "[(44, 10)]\n",
      "Edges removed: \n",
      "[(20, 152)]\n",
      "Edges removed: \n",
      "[(107, 8)]\n",
      "Edges removed: \n",
      "[(133, 9)]\n",
      "Edges removed: \n",
      "[(5, 8)]\n",
      "Edges removed: \n",
      "[(21, 93)]\n",
      "Edges removed: \n",
      "[(15, 86)]\n",
      "Edges removed: \n",
      "[(191, 50)]\n",
      "Edges removed: \n",
      "[(17, 188)]\n",
      "Edges removed: \n",
      "[(11, 108)]\n",
      "Edges removed: \n",
      "[(158, 21)]\n",
      "Edges removed: \n",
      "[(9, 100)]\n",
      "Edges removed: \n",
      "[(149, 15)]\n",
      "Edges removed: \n",
      "[(148, 2)]\n",
      "Edges removed: \n",
      "[(29, 85)]\n",
      "Edges removed: \n",
      "[(66, 29)]\n",
      "Edges removed: \n",
      "[(23, 76)]\n",
      "Edges removed: \n",
      "[(12, 24)]\n",
      "Edges removed: \n",
      "[(11, 32)]\n",
      "Edges removed: \n",
      "[(75, 25)]\n",
      "Edges removed: \n",
      "[(25, 140)]\n",
      "Edges removed: \n",
      "[(23, 24)]\n",
      "Edges removed: \n",
      "[(13, 24)]\n",
      "Edges removed: \n",
      "[(19, 91)]\n",
      "Edges removed: \n",
      "[(13, 179)]\n",
      "Edges removed: \n",
      "[(82, 32)]\n",
      "Edges removed: \n",
      "[(175, 23)]\n",
      "Edges removed: \n",
      "[(7, 19)]\n",
      "Edges removed: \n",
      "[(3, 19)]\n",
      "Edges removed: \n",
      "[(132, 23)]\n",
      "Edges removed: \n",
      "[(176, 23)]\n",
      "Edges removed: \n",
      "[(3, 51)]\n",
      "Edges removed: \n",
      "[(17, 22)]\n",
      "Edges removed: \n",
      "[(158, 17)]\n",
      "Edges removed: \n",
      "[(10, 187)]\n",
      "Edges removed: \n",
      "[(10, 74)]\n",
      "Edges removed: \n",
      "[(112, 10)]\n",
      "Edges removed: \n",
      "[(3, 123)]\n",
      "Edges removed: \n",
      "[(126, 22)]\n",
      "Edges removed: \n",
      "[(15, 28)]\n",
      "Edges removed: \n",
      "[(15, 133)]\n",
      "Edges removed: \n",
      "[(4, 73)]\n",
      "Edges removed: \n",
      "[(194, 13)]\n",
      "Edges removed: \n",
      "[(0, 47)]\n",
      "Edges removed: \n",
      "[(180, 12)]\n",
      "Edges removed: \n",
      "[(9, 185)]\n",
      "Edges removed: \n",
      "[(68, 21)]\n",
      "Edges removed: \n",
      "[(25, 24)]\n",
      "Edges removed: \n",
      "[(21, 119)]\n",
      "Edges removed: \n",
      "[(80, 21)]\n",
      "Edges removed: \n",
      "[(33, 96)]\n",
      "Edges removed: \n",
      "[(22, 177)]\n",
      "Edges removed: \n",
      "[(19, 33)]\n",
      "Edges removed: \n",
      "[(181, 47)]\n",
      "Edges removed: \n",
      "[(160, 27)]\n",
      "Edges removed: \n",
      "[(3, 110)]\n",
      "Edges removed: \n",
      "[(122, 3)]\n",
      "Edges removed: \n",
      "[(6, 50)]\n",
      "Edges removed: \n",
      "[(6, 94)]\n",
      "Edges removed: \n",
      "[(3, 29)]\n",
      "Edges removed: \n",
      "[(6, 29)]\n",
      "Edges removed: \n",
      "[(6, 120)]\n",
      "Edges removed: \n",
      "[(114, 47)]\n",
      "Edges removed: \n",
      "[(105, 47)]\n",
      "Edges removed: \n",
      "[(4, 88)]\n",
      "Edges removed: \n",
      "[(4, 119)]\n",
      "Edges removed: \n",
      "[(183, 18)]\n",
      "Edges removed: \n",
      "[(192, 18)]\n",
      "Edges removed: \n",
      "[(12, 81)]\n",
      "Edges removed: \n",
      "[(128, 27)]\n",
      "Edges removed: \n",
      "[(185, 24)]\n",
      "Edges removed: \n",
      "[(12, 68)]\n",
      "Edges removed: \n",
      "[(148, 4)]\n",
      "Edges removed: \n",
      "[(93, 4)]\n",
      "Edges removed: \n",
      "[(134, 11)]\n",
      "Edges removed: \n",
      "[(119, 11)]\n",
      "Edges removed: \n",
      "[(7, 173)]\n",
      "Edges removed: \n",
      "[(5, 112)]\n",
      "Edges removed: \n",
      "[(12, 199)]\n",
      "Edges removed: \n",
      "[(12, 75)]\n",
      "Edges removed: \n",
      "[(103, 12)]\n",
      "Edges removed: \n",
      "[(161, 6)]\n",
      "Edges removed: \n",
      "[(55, 6)]\n",
      "Edges removed: \n",
      "[(0, 148)]\n",
      "Edges removed: \n",
      "[(86, 20)]\n",
      "Edges removed: \n",
      "[(68, 47)]\n",
      "Edges removed: \n",
      "[(119, 47)]\n",
      "Edges removed: \n",
      "[(2, 68)]\n",
      "Edges removed: \n",
      "[(110, 24)]\n",
      "Edges removed: \n",
      "[(105, 24)]\n",
      "Edges removed: \n",
      "[(9, 181)]\n",
      "Edges removed: \n",
      "[(155, 11)]\n",
      "Edges removed: \n",
      "[(76, 13)]\n",
      "Edges removed: \n",
      "[(107, 13)]\n",
      "Edges removed: \n",
      "[(71, 13)]\n",
      "Edges removed: \n",
      "[(6, 130)]\n",
      "Edges removed: \n",
      "[(6, 95)]\n",
      "Edges removed: \n",
      "[(2, 162)]\n",
      "Edges removed: \n",
      "[(103, 8)]\n",
      "Edges removed: \n",
      "[(130, 20)]\n",
      "Edges removed: \n",
      "[(87, 20)]\n",
      "Edges removed: \n",
      "[(6, 83)]\n",
      "Edges removed: \n",
      "[(171, 21)]\n",
      "Edges removed: \n",
      "[(1, 35)]\n",
      "Edges removed: \n",
      "[(159, 5)]\n",
      "Edges removed: \n",
      "[(11, 173)]\n",
      "Edges removed: \n",
      "[(160, 18)]\n",
      "Edges removed: \n",
      "[(150, 9)]\n",
      "Edges removed: \n",
      "[(7, 99)]\n",
      "Edges removed: \n",
      "[(23, 19)]\n",
      "Edges removed: \n",
      "[(23, 140)]\n",
      "Edges removed: \n",
      "[(10, 19)]\n",
      "Edges removed: \n",
      "[(134, 19)]\n",
      "Edges removed: \n",
      "[(76, 12)]\n",
      "Edges removed: \n",
      "[(11, 160)]\n",
      "Edges removed: \n",
      "[(88, 23)]\n",
      "Edges removed: \n",
      "[(11, 115)]\n",
      "Edges removed: \n",
      "[(11, 188)]\n",
      "Edges removed: \n",
      "[(115, 23)]\n",
      "Edges removed: \n",
      "[(188, 23)]\n",
      "Edges removed: \n",
      "[(125, 11)]\n",
      "Edges removed: \n",
      "[(181, 11)]\n",
      "Edges removed: \n",
      "[(89, 11)]\n",
      "Edges removed: \n",
      "[(6, 171)]\n",
      "Edges removed: \n",
      "[(121, 18)]\n",
      "Edges removed: \n",
      "[(166, 18)]\n",
      "Edges removed: \n",
      "[(42, 18)]\n",
      "Edges removed: \n",
      "[(0, 181)]\n",
      "Edges removed: \n",
      "[(86, 6)]\n",
      "Edges removed: \n",
      "[(71, 6)]\n",
      "Edges removed: \n",
      "[(130, 13)]\n",
      "Edges removed: \n",
      "[(103, 13)]\n",
      "Edges removed: \n",
      "[(134, 13)]\n",
      "Edges removed: \n",
      "[(152, 31)]\n",
      "Edges removed: \n",
      "[(2, 103)]\n",
      "Edges removed: \n",
      "[(199, 31)]\n",
      "Edges removed: \n",
      "[(0, 149)]\n",
      "Edges removed: \n",
      "[(0, 116)]\n",
      "Edges removed: \n",
      "[(0, 140)]\n",
      "Edges removed: \n",
      "[(0, 93)]\n",
      "Edges removed: \n",
      "[(129, 28)]\n",
      "Edges removed: \n",
      "[(2, 152)]\n",
      "Edges removed: \n",
      "[(2, 199)]\n",
      "Edges removed: \n",
      "[(2, 94)]\n",
      "Edges removed: \n",
      "[(2, 110)]\n",
      "Edges removed: \n",
      "[(44, 6)]\n",
      "Edges removed: \n",
      "[(2, 189)]\n",
      "Edges removed: \n",
      "[(2, 42)]\n",
      "Edges removed: \n",
      "[(2, 122)]\n",
      "Finished\n",
      "Total time 621.3434615135193\n",
      "Distortion 0.43390191897654584\n"
     ]
    }
   ],
   "source": [
    "G_aux = anonimize(G1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Edges removed: \n",
      "[(35, 75)]\n",
      "Edges removed: \n",
      "[(51, 167)]\n",
      "Edges removed: \n",
      "[(35, 128)]\n",
      "Edges removed: \n",
      "[(21, 195)]\n",
      "Edges removed: \n",
      "[(58, 178)]\n",
      "Edges removed: \n",
      "[(15, 151)]\n",
      "Edges removed: \n",
      "[(22, 97)]\n",
      "Edges removed: \n",
      "[(2, 169)]\n",
      "Edges removed: \n",
      "[(6, 124)]\n",
      "Edges removed: \n",
      "[(47, 198)]\n",
      "Edges removed: \n",
      "[(30, 186)]\n",
      "Edges removed: \n",
      "[(41, 120)]\n",
      "Edges removed: \n",
      "[(4, 110)]\n",
      "Edges removed: \n",
      "[(12, 160)]\n",
      "Edges removed: \n",
      "[(43, 28)]\n",
      "Edges removed: \n",
      "[(6, 144)]\n",
      "Edges removed: \n",
      "[(18, 168)]\n",
      "Edges removed: \n",
      "[(19, 134)]\n",
      "Edges removed: \n",
      "[(56, 186)]\n",
      "Edges removed: \n",
      "[(11, 34)]\n",
      "Edges removed: \n",
      "[(4, 135)]\n",
      "Edges removed: \n",
      "[(27, 7)]\n",
      "Edges removed: \n",
      "[(84, 134)]\n",
      "Edges removed: \n",
      "[(15, 149)]\n",
      "Edges removed: \n",
      "[(1, 35)]\n",
      "Edges removed: \n",
      "[(46, 160)]\n",
      "Edges removed: \n",
      "[(71, 121)]\n",
      "Edges removed: \n",
      "[(33, 68)]\n",
      "Edges removed: \n",
      "[(68, 165)]\n",
      "Edges removed: \n",
      "[(0, 198)]\n",
      "Edges removed: \n",
      "[(74, 66)]\n",
      "Edges removed: \n",
      "[(41, 118)]\n",
      "Edges removed: \n",
      "[(4, 9)]\n",
      "Edges removed: \n",
      "[(5, 86)]\n",
      "Edges removed: \n",
      "[(83, 167)]\n",
      "Edges removed: \n",
      "[(67, 167)]\n",
      "Edges removed: \n",
      "[(0, 178)]\n",
      "Edges removed: \n",
      "[(3, 186)]\n",
      "Edges removed: \n",
      "[(9, 124)]\n",
      "Edges removed: \n",
      "[(17, 167)]\n",
      "Edges removed: \n",
      "[(57, 167)]\n",
      "Edges removed: \n",
      "[(3, 167)]\n",
      "Edges removed: \n",
      "[(10, 66)]\n",
      "Edges removed: \n",
      "[(47, 197)]\n",
      "Edges removed: \n",
      "[(40, 117)]\n",
      "Edges removed: \n",
      "[(33, 117)]\n",
      "Edges removed: \n",
      "[(8, 117)]\n",
      "Edges removed: \n",
      "[(8, 170)]\n",
      "Edges removed: \n",
      "[(34, 170)]\n",
      "Edges removed: \n",
      "[(20, 59)]\n",
      "Edges removed: \n",
      "[(6, 110)]\n",
      "Edges removed: \n",
      "[(6, 178)]\n",
      "Edges removed: \n",
      "[(6, 105)]\n",
      "Edges removed: \n",
      "[(76, 178)]\n",
      "Edges removed: \n",
      "[(52, 56)]\n",
      "Edges removed: \n",
      "[(40, 36)]\n",
      "Edges removed: \n",
      "[(9, 41)]\n",
      "Edges removed: \n",
      "[(19, 77)]\n",
      "Edges removed: \n",
      "[(7, 21)]\n",
      "Edges removed: \n",
      "[(5, 55)]\n",
      "Edges removed: \n",
      "[(5, 183)]\n",
      "Edges removed: \n",
      "[(28, 168)]\n",
      "Edges removed: \n",
      "[(4, 168)]\n",
      "Edges removed: \n",
      "[(15, 21)]\n",
      "Edges removed: \n",
      "[(21, 184)]\n",
      "Edges removed: \n",
      "[(21, 115)]\n",
      "Edges removed: \n",
      "[(0, 183)]\n",
      "Edges removed: \n",
      "[(93, 167)]\n",
      "Edges removed: \n",
      "[(49, 72)]\n",
      "Edges removed: \n",
      "[(92, 99)]\n",
      "Edges removed: \n",
      "[(18, 165)]\n",
      "Edges removed: \n",
      "[(17, 105)]\n",
      "Edges removed: \n",
      "[(17, 21)]\n",
      "Edges removed: \n",
      "[(20, 45)]\n",
      "Edges removed: \n",
      "[(21, 105)]\n",
      "Edges removed: \n",
      "[(15, 185)]\n",
      "Edges removed: \n",
      "[(6, 71)]\n",
      "Edges removed: \n",
      "[(71, 55)]\n",
      "Edges removed: \n",
      "[(46, 71)]\n",
      "Edges removed: \n",
      "[(19, 33)]\n",
      "Edges removed: \n",
      "[(152, 125)]\n",
      "Edges removed: \n",
      "[(13, 95)]\n",
      "Edges removed: \n",
      "[(7, 150)]\n",
      "Edges removed: \n",
      "[(7, 52)]\n",
      "Edges removed: \n",
      "[(84, 168)]\n",
      "Edges removed: \n",
      "[(7, 104)]\n",
      "Edges removed: \n",
      "[(34, 184)]\n",
      "Edges removed: \n",
      "[(34, 115)]\n",
      "Edges removed: \n",
      "[(7, 129)]\n",
      "Edges removed: \n",
      "[(18, 117)]\n",
      "Edges removed: \n",
      "[(18, 141)]\n",
      "Edges removed: \n",
      "[(1, 119)]\n",
      "Edges removed: \n",
      "[(2, 156)]\n",
      "Edges removed: \n",
      "[(33, 85)]\n",
      "Edges removed: \n",
      "[(12, 137)]\n",
      "Edges removed: \n",
      "[(22, 105)]\n",
      "Edges removed: \n",
      "[(36, 105)]\n",
      "Edges removed: \n",
      "[(49, 101)]\n",
      "Edges removed: \n",
      "[(10, 125)]\n",
      "Edges removed: \n",
      "[(29, 76)]\n",
      "Edges removed: \n",
      "[(17, 184)]\n",
      "Edges removed: \n",
      "[(5, 168)]\n",
      "Edges removed: \n",
      "[(6, 88)]\n",
      "Edges removed: \n",
      "[(15, 80)]\n",
      "Edges removed: \n",
      "[(15, 170)]\n",
      "Edges removed: \n",
      "[(19, 23)]\n",
      "Edges removed: \n",
      "[(19, 73)]\n",
      "Edges removed: \n",
      "[(26, 73)]\n",
      "Edges removed: \n",
      "[(6, 95)]\n",
      "Edges removed: \n",
      "[(29, 143)]\n",
      "Edges removed: \n",
      "[(15, 28)]\n",
      "Edges removed: \n",
      "[(50, 184)]\n",
      "Edges removed: \n",
      "[(23, 140)]\n",
      "Edges removed: \n",
      "[(52, 101)]\n",
      "Edges removed: \n",
      "[(17, 164)]\n",
      "Edges removed: \n",
      "[(17, 95)]\n",
      "Edges removed: \n",
      "[(58, 129)]\n",
      "Edges removed: \n",
      "[(75, 132)]\n",
      "Edges removed: \n",
      "[(81, 132)]\n",
      "Edges removed: \n",
      "[(56, 94)]\n",
      "Edges removed: \n",
      "[(15, 93)]\n",
      "Edges removed: \n",
      "[(25, 190)]\n",
      "Edges removed: \n",
      "[(80, 137)]\n",
      "Edges removed: \n",
      "[(44, 175)]\n",
      "Edges removed: \n",
      "[(51, 140)]\n",
      "Edges removed: \n",
      "[(19, 140)]\n",
      "Edges removed: \n",
      "[(14, 71)]\n",
      "Edges removed: \n",
      "[(13, 123)]\n",
      "Edges removed: \n",
      "[(13, 39)]\n",
      "Edges removed: \n",
      "[(39, 172)]\n",
      "Edges removed: \n",
      "[(37, 68)]\n",
      "Edges removed: \n",
      "[(10, 93)]\n",
      "Edges removed: \n",
      "[(4, 102)]\n",
      "Edges removed: \n",
      "[(4, 129)]\n",
      "Edges removed: \n",
      "[(14, 129)]\n",
      "Edges removed: \n",
      "[(34, 2)]\n",
      "Edges removed: \n",
      "[(25, 95)]\n",
      "Edges removed: \n",
      "[(64, 53)]\n",
      "Edges removed: \n",
      "[(31, 102)]\n",
      "Edges removed: \n",
      "[(32, 76)]\n",
      "Edges removed: \n",
      "[(76, 96)]\n",
      "Edges removed: \n",
      "[(79, 184)]\n",
      "Edges removed: \n",
      "[(26, 33)]\n",
      "Edges removed: \n",
      "[(80, 184)]\n",
      "Edges removed: \n",
      "[(75, 106)]\n",
      "Edges removed: \n",
      "[(79, 90)]\n",
      "Edges removed: \n",
      "[(48, 107)]\n",
      "Edges removed: \n",
      "[(25, 48)]\n",
      "Edges removed: \n",
      "[(48, 53)]\n",
      "Edges removed: \n",
      "[(48, 63)]\n",
      "Edges removed: \n",
      "[(48, 2)]\n",
      "Edges removed: \n",
      "[(48, 31)]\n",
      "Edges removed: \n",
      "[(2, 89)]\n",
      "Edges removed: \n",
      "[(0, 107)]\n",
      "Edges removed: \n",
      "[(21, 107)]\n",
      "Edges removed: \n",
      "[(0, 83)]\n",
      "Edges removed: \n",
      "[(28, 87)]\n",
      "Edges removed: \n",
      "[(37, 87)]\n",
      "Edges removed: \n",
      "[(57, 101)]\n",
      "Edges removed: \n",
      "[(3, 37)]\n",
      "Edges removed: \n",
      "[(29, 37)]\n",
      "Edges removed: \n",
      "[(22, 29)]\n",
      "Edges removed: \n",
      "[(28, 101)]\n",
      "Edges removed: \n",
      "[(14, 88)]\n",
      "Edges removed: \n",
      "[(14, 59)]\n",
      "Edges removed: \n",
      "[(59, 125)]\n",
      "Edges removed: \n",
      "[(18, 51)]\n",
      "Edges removed: \n",
      "[(0, 41)]\n",
      "Edges removed: \n",
      "[(21, 67)]\n",
      "Edges removed: \n",
      "[(5, 53)]\n",
      "Edges removed: \n",
      "[(15, 58)]\n",
      "Edges removed: \n",
      "[(24, 84)]\n",
      "Edges removed: \n",
      "[(24, 39)]\n",
      "Edges removed: \n",
      "[(15, 22)]\n",
      "Edges removed: \n",
      "[(15, 19)]\n",
      "Edges removed: \n",
      "[(29, 66)]\n",
      "Edges removed: \n",
      "[(29, 72)]\n",
      "Edges removed: \n",
      "[(20, 152)]\n",
      "Edges removed: \n",
      "[(6, 83)]\n",
      "Edges removed: \n",
      "[(6, 46)]\n",
      "Edges removed: \n",
      "[(55, 106)]\n",
      "Edges removed: \n",
      "[(6, 76)]\n",
      "Edges removed: \n",
      "[(19, 66)]\n",
      "Edges removed: \n",
      "[(21, 63)]\n",
      "Edges removed: \n",
      "[(32, 84)]\n",
      "Edges removed: \n",
      "[(37, 62)]\n",
      "Edges removed: \n",
      "[(21, 54)]\n",
      "Edges removed: \n",
      "[(5, 125)]\n",
      "Edges removed: \n",
      "[(45, 63)]\n",
      "Edges removed: \n",
      "[(5, 63)]\n",
      "Edges removed: \n",
      "[(88, 125)]\n",
      "Edges removed: \n",
      "[(45, 65)]\n",
      "Edges removed: \n",
      "[(44, 123)]\n",
      "Edges removed: \n",
      "[(69, 97)]\n",
      "Edges removed: \n",
      "[(71, 75)]\n",
      "Edges removed: \n",
      "[(13, 82)]\n",
      "Edges removed: \n",
      "[(79, 82)]\n",
      "Edges removed: \n",
      "[(3, 79)]\n",
      "Edges removed: \n",
      "[(5, 41)]\n",
      "Edges removed: \n",
      "[(2, 152)]\n",
      "Edges removed: \n",
      "[(75, 99)]\n",
      "Edges removed: \n",
      "[(60, 184)]\n",
      "Edges removed: \n",
      "[(22, 60)]\n",
      "Edges removed: \n",
      "[(79, 81)]\n",
      "Edges removed: \n",
      "[(46, 101)]\n",
      "Edges removed: \n",
      "[(19, 100)]\n",
      "Edges removed: \n",
      "[(50, 100)]\n",
      "Edges removed: \n",
      "[(6, 60)]\n",
      "Edges removed: \n",
      "[(6, 50)]\n",
      "Edges removed: \n",
      "[(12, 180)]\n",
      "Edges removed: \n",
      "[(28, 68)]\n",
      "Edges removed: \n",
      "[(5, 69)]\n",
      "Edges removed: \n",
      "[(5, 56)]\n",
      "Edges removed: \n",
      "[(5, 13)]\n",
      "Edges removed: \n",
      "[(8, 64)]\n",
      "Edges removed: \n",
      "[(12, 97)]\n",
      "Edges removed: \n",
      "[(12, 68)]\n",
      "Edges removed: \n",
      "[(51, 80)]\n",
      "Edges removed: \n",
      "[(29, 80)]\n",
      "Edges removed: \n",
      "[(37, 82)]\n",
      "Edges removed: \n",
      "[(29, 38)]\n",
      "Edges removed: \n",
      "[(1, 74)]\n",
      "Edges removed: \n",
      "[(14, 92)]\n",
      "Edges removed: \n",
      "[(6, 89)]\n",
      "Edges removed: \n",
      "[(6, 19)]\n",
      "Edges removed: \n",
      "[(55, 57)]\n",
      "Edges removed: \n",
      "[(44, 89)]\n",
      "Edges removed: \n",
      "[(10, 85)]\n",
      "Edges removed: \n",
      "[(51, 89)]\n",
      "Edges removed: \n",
      "[(22, 184)]\n",
      "Edges removed: \n",
      "[(21, 80)]\n",
      "Edges removed: \n",
      "[(21, 108)]\n",
      "Edges removed: \n",
      "[(28, 59)]\n",
      "Edges removed: \n",
      "[(62, 59)]\n",
      "Edges removed: \n",
      "[(1, 92)]\n",
      "Edges removed: \n",
      "[(24, 110)]\n",
      "Edges removed: \n",
      "[(59, 110)]\n",
      "Edges removed: \n",
      "[(12, 56)]\n",
      "Edges removed: \n",
      "[(46, 48)]\n",
      "Edges removed: \n",
      "[(67, 152)]\n",
      "Edges removed: \n",
      "[(12, 14)]\n",
      "Edges removed: \n",
      "[(7, 105)]\n",
      "Edges removed: \n",
      "[(7, 84)]\n",
      "Edges removed: \n",
      "[(36, 84)]\n",
      "Edges removed: \n",
      "[(56, 79)]\n",
      "Edges removed: \n",
      "[(56, 80)]\n",
      "Edges removed: \n",
      "[(6, 44)]\n",
      "Edges removed: \n",
      "[(50, 29)]\n",
      "Edges removed: \n",
      "[(29, 85)]\n",
      "Edges removed: \n",
      "[(10, 121)]\n",
      "Edges removed: \n",
      "[(7, 78)]\n",
      "Edges removed: \n",
      "[(55, 152)]\n",
      "Edges removed: \n",
      "[(39, 87)]\n",
      "Edges removed: \n",
      "[(19, 64)]\n",
      "Edges removed: \n",
      "[(19, 10)]\n",
      "Edges removed: \n",
      "[(17, 36)]\n",
      "Edges removed: \n",
      "[(36, 102)]\n",
      "Edges removed: \n",
      "[(13, 71)]\n",
      "Edges removed: \n",
      "[(27, 81)]\n",
      "Edges removed: \n",
      "[(50, 69)]\n",
      "Edges removed: \n",
      "[(28, 172)]\n",
      "Edges removed: \n",
      "[(25, 80)]\n",
      "Edges removed: \n",
      "[(49, 92)]\n",
      "Edges removed: \n",
      "[(17, 85)]\n",
      "Edges removed: \n",
      "[(17, 22)]\n",
      "Edges removed: \n",
      "[(17, 42)]\n",
      "Edges removed: \n",
      "[(17, 108)]\n",
      "Edges removed: \n",
      "[(17, 2)]\n",
      "Edges removed: \n",
      "[(9, 108)]\n",
      "Edges removed: \n",
      "[(9, 181)]\n",
      "Edges removed: \n",
      "[(9, 25)]\n",
      "Edges removed: \n",
      "[(29, 108)]\n",
      "Edges removed: \n",
      "[(9, 85)]\n",
      "Edges removed: \n",
      "[(5, 33)]\n",
      "Edges removed: \n",
      "[(132, 97)]\n",
      "Edges removed: \n",
      "[(7, 196)]\n",
      "Edges removed: \n",
      "[(182, 108)]\n",
      "Edges removed: \n",
      "[(181, 52)]\n",
      "Edges removed: \n",
      "[(56, 84)]\n",
      "Edges removed: \n",
      "[(54, 84)]\n",
      "Edges removed: \n",
      "[(83, 84)]\n",
      "Edges removed: \n",
      "[(81, 84)]\n",
      "Edges removed: \n",
      "[(8, 107)]\n",
      "Edges removed: \n",
      "[(19, 142)]\n",
      "Edges removed: \n",
      "[(19, 57)]\n",
      "Edges removed: \n",
      "[(13, 90)]\n",
      "Edges removed: \n",
      "[(73, 75)]\n",
      "Edges removed: \n",
      "[(1, 80)]\n",
      "Edges removed: \n",
      "[(59, 26)]\n",
      "Edges removed: \n",
      "[(42, 16)]\n",
      "Edges removed: \n",
      "[(21, 106)]\n",
      "Edges removed: \n",
      "[(21, 70)]\n",
      "Edges removed: \n",
      "[(10, 126)]\n",
      "Edges removed: \n",
      "[(10, 62)]\n",
      "Edges removed: \n",
      "[(42, 21)]\n",
      "Edges removed: \n",
      "[(25, 26)]\n",
      "Edges removed: \n",
      "[(4, 106)]\n",
      "Edges removed: \n",
      "[(50, 85)]\n",
      "Edges removed: \n",
      "[(48, 62)]\n",
      "Edges removed: \n",
      "[(39, 149)]\n",
      "Edges removed: \n",
      "[(148, 97)]\n",
      "Edges removed: \n",
      "[(59, 38)]\n",
      "Edges removed: \n",
      "[(7, 67)]\n",
      "Edges removed: \n",
      "[(13, 30)]\n",
      "Edges removed: \n",
      "[(13, 102)]\n",
      "Edges removed: \n",
      "[(6, 161)]\n",
      "Edges removed: \n",
      "[(25, 50)]\n",
      "Edges removed: \n",
      "[(1, 163)]\n",
      "Edges removed: \n",
      "[(155, 50)]\n",
      "Edges removed: \n",
      "[(21, 155)]\n",
      "Edges removed: \n",
      "[(0, 47)]\n",
      "Edges removed: \n",
      "[(19, 172)]\n",
      "Edges removed: \n",
      "[(23, 139)]\n",
      "Edges removed: \n",
      "[(15, 94)]\n",
      "Edges removed: \n",
      "[(24, 59)]\n",
      "Edges removed: \n",
      "[(19, 85)]\n",
      "Edges removed: \n",
      "[(9, 100)]\n",
      "Edges removed: \n",
      "[(2, 87)]\n",
      "Edges removed: \n",
      "[(28, 56)]\n",
      "Edges removed: \n",
      "[(40, 152)]\n",
      "Edges removed: \n",
      "[(28, 152)]\n",
      "Edges removed: \n",
      "[(28, 37)]\n",
      "Edges removed: \n",
      "[(21, 45)]\n",
      "Edges removed: \n",
      "[(37, 81)]\n",
      "Edges removed: \n",
      "[(104, 9)]\n",
      "Edges removed: \n",
      "[(2, 44)]\n",
      "Edges removed: \n",
      "[(89, 50)]\n",
      "Edges removed: \n",
      "[(27, 70)]\n",
      "Edges removed: \n",
      "[(93, 56)]\n",
      "Edges removed: \n",
      "[(185, 45)]\n",
      "Edges removed: \n",
      "[(5, 112)]\n",
      "Edges removed: \n",
      "[(23, 170)]\n",
      "Edges removed: \n",
      "[(21, 25)]\n",
      "Edges removed: \n",
      "[(1, 57)]\n",
      "Edges removed: \n",
      "[(1, 9)]\n",
      "Edges removed: \n",
      "[(0, 156)]\n",
      "Edges removed: \n",
      "[(191, 85)]\n",
      "Edges removed: \n",
      "[(41, 88)]\n",
      "Edges removed: \n",
      "[(42, 181)]\n",
      "Edges removed: \n",
      "[(92, 62)]\n",
      "Edges removed: \n",
      "[(10, 174)]\n",
      "Edges removed: \n",
      "[(3, 29)]\n",
      "Edges removed: \n",
      "[(73, 31)]\n",
      "Edges removed: \n",
      "[(41, 104)]\n",
      "Edges removed: \n",
      "[(5, 107)]\n",
      "Edges removed: \n",
      "[(21, 47)]\n",
      "Edges removed: \n",
      "[(28, 51)]\n",
      "Edges removed: \n",
      "[(28, 199)]\n",
      "Edges removed: \n",
      "[(28, 70)]\n",
      "Edges removed: \n",
      "[(20, 35)]\n",
      "Edges removed: \n",
      "[(159, 5)]\n",
      "Edges removed: \n",
      "[(129, 28)]\n",
      "Edges removed: \n",
      "[(56, 40)]\n",
      "Edges removed: \n",
      "[(172, 71)]\n",
      "Edges removed: \n",
      "[(2, 134)]\n",
      "Edges removed: \n",
      "[(4, 22)]\n",
      "Edges removed: \n",
      "[(42, 50)]\n",
      "Edges removed: \n",
      "[(4, 111)]\n",
      "Edges removed: \n",
      "[(1, 112)]\n",
      "Edges removed: \n",
      "[(10, 41)]\n",
      "Edges removed: \n",
      "[(21, 28)]\n",
      "Edges removed: \n",
      "[(19, 119)]\n",
      "Edges removed: \n",
      "[(19, 28)]\n",
      "Edges removed: \n",
      "[(181, 47)]\n",
      "Edges removed: \n",
      "[(93, 28)]\n",
      "Finished\n",
      "Total time 1512.610279083252\n",
      "Distortion 0.2812018489984592\n"
     ]
    }
   ],
   "source": [
    "G_san = anonimize(G2)"
   ]
  },
  {
   "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'200 vs 1298'"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "str(G_aux.number_of_nodes())+  ' vs '  +  str(G_san.number_of_edges())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "G_a1, G_a2 = edge_split(G_aux, 0.25) # aux \n",
    "G_s1, G_s2 = edge_split(G_san, 0.25) # sanitized"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "'''\n",
    " Get the histogram of degree distribution for an node 'n' in a graph 'G' \n",
    "     bins - the number of bins in the deg. distribution \n",
    "     size - size of the historgram \n",
    "'''\n",
    "def deg_dist(G, n, bins, size):\n",
    "    feature_set = [0 for i in range(2*bins)]\n",
    "    _1hop = G[n]\n",
    "    _1hop = G.degree(_1hop).values()\n",
    "    for h in _1hop:\n",
    "        if h < bins*size:\n",
    "            feature_set[int(h/size)] += 1\n",
    "    _prev = set(nx.single_source_shortest_path_length(G, n, cutoff=1).keys())\n",
    "    _2hop = set(nx.single_source_shortest_path_length(G, n, cutoff=2).keys())\n",
    "    _2hop = _2hop - _prev\n",
    "    _2hop = G.degree(_2hop).values()\n",
    "    for h in _2hop:\n",
    "        if h < bins*size:\n",
    "            feature_set[bins+int(h/size)] += 1\n",
    "    return feature_set\n",
    "\n",
    "\n",
    "''' Get the feature vector for egde e1 in graph G1\n",
    "     and e2 in graph G2.\n",
    "     1hop, 2hop deg distributions\n",
    "     + 4 Silhoutte coefficients\n",
    "'''    \n",
    "def e_feature(G1, e1, G2, e2,  bins = 21, size = 50):\n",
    "    feature_set = [0 for i in range(8*bins)]\n",
    "    for j in [0, 1]:\n",
    "        f_set = deg_dist(G1, e1[j], bins, size)\n",
    "        for i in range(len(f_set)):\n",
    "            feature_set[2*bins*j + i] += f_set[i]\n",
    "    for j in [0, 1]:\n",
    "        f_set = deg_dist(G2, e2[j], bins, size)\n",
    "        for i in range(len(f_set)):\n",
    "            feature_set[4*bins+2*bins*j + i] += f_set[i]\n",
    "    \n",
    "    # 4 Silhouette Coefficients \n",
    "    e1_deg0 = G1.degree(e1[0])\n",
    "    e1_deg1 = G1.degree(e1[1])\n",
    "    \n",
    "    e2_deg0 = G2.degree(e2[0])\n",
    "    e2_deg1 = G2.degree(e2[1])\n",
    "    \n",
    "    feature_set.append(abs(e1_deg0-e2_deg0)/max(e1_deg0, e2_deg0, 1))\n",
    "    feature_set.append(abs(e1_deg1-e2_deg1)/max(e1_deg1, e2_deg1, 1))\n",
    "    feature_set.append(abs(e1_deg0-e2_deg1)/max(e1_deg0, e2_deg1, 1))\n",
    "    feature_set.append(abs(e1_deg1-e2_deg0)/max(e1_deg1, e2_deg0, 1))\n",
    "    \n",
    "    return feature_set\n",
    "\n",
    "'''\n",
    "    Transform graph into feature vector dataset\n",
    "    G1, G2 - graphs for comparison\n",
    "'''\n",
    "def fill_set(G1, G2, data, labels, n_max=None):\n",
    "    indx = 0\n",
    "    total = len(G1.nodes())\n",
    "    # report every 10 % \n",
    "    part = total//10\n",
    "    # Add all intersections first: \n",
    "    print('Start adding positive examples')\n",
    "    common = set(G1.edges()) & set(G2.edges())\n",
    "    if n_max is None: \n",
    "        n_max = len(common)\n",
    "    common = random.sample(common, n_max)\n",
    "    indx = 0\n",
    "    for i in common: \n",
    "        data.append(e_feature(G1,i, G2, i))\n",
    "        labels.append(1)\n",
    "        indx += 1\n",
    "        if indx % int(len(common)/10) ==0:\n",
    "            print(\"Finished : \" + str(indx/len(common)))\n",
    "        \n",
    "    g1 = set(G1.edges()) - set(common)\n",
    "    g2 = set(G2.edges()) - set(common)\n",
    "    \n",
    "    # How many false results to add ? - \n",
    "    # The same size as common\n",
    "    # TODO: \n",
    "    print('Start adding false examples')\n",
    "    if len(g1) < n_max:\n",
    "        n_max = min(len(g1), len(g2))\n",
    "    g1 = random.sample(g1, n_max)\n",
    "    g2 = random.sample(g2, n_max)\n",
    "    for i in range(len(g1)):\n",
    "        data.append(e_feature(G1, g1[i], G2, g2[i]))\n",
    "        labels.append(0)\n",
    "        if i % int(len(g1)/10) ==0:\n",
    "            print(\"Finished : \" + str(i/len(g1)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.09558823529411764\n",
      "Finished : 0.19117647058823528\n",
      "Finished : 0.2867647058823529\n",
      "Finished : 0.38235294117647056\n",
      "Finished : 0.47794117647058826\n",
      "Finished : 0.5735294117647058\n",
      "Finished : 0.6691176470588235\n",
      "Finished : 0.7647058823529411\n",
      "Finished : 0.8602941176470589\n",
      "Finished : 0.9558823529411765\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.09090909090909091\n",
      "Finished : 0.18181818181818182\n",
      "Finished : 0.2727272727272727\n",
      "Finished : 0.36363636363636365\n",
      "Finished : 0.45454545454545453\n",
      "Finished : 0.5454545454545454\n",
      "Finished : 0.6363636363636364\n",
      "Finished : 0.7272727272727273\n",
      "Finished : 0.8181818181818182\n",
      "Finished : 0.9090909090909091\n",
      "Aux finished 0.08351683616638184 sec\n",
      "Start adding positive examples\n",
      "Finished : 0.09606986899563319\n",
      "Finished : 0.19213973799126638\n",
      "Finished : 0.28820960698689957\n",
      "Finished : 0.38427947598253276\n",
      "Finished : 0.48034934497816595\n",
      "Finished : 0.5764192139737991\n",
      "Finished : 0.6724890829694323\n",
      "Finished : 0.7685589519650655\n",
      "Finished : 0.8646288209606987\n",
      "Finished : 0.9606986899563319\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.09523809523809523\n",
      "Finished : 0.19047619047619047\n",
      "Finished : 0.2857142857142857\n",
      "Finished : 0.38095238095238093\n",
      "Finished : 0.47619047619047616\n",
      "Finished : 0.5714285714285714\n",
      "Finished : 0.6666666666666666\n",
      "Finished : 0.7619047619047619\n",
      "Finished : 0.8571428571428571\n",
      "Finished : 0.9523809523809523\n",
      "San finished 0.1372368335723877 sec\n"
     ]
    }
   ],
   "source": [
    "# Generating train dataset: \n",
    "train_set = []\n",
    "train_labels = []\n",
    "s = time()\n",
    "print(\"Dataset is generating ... \")\n",
    "fill_set(G_a1, G_a2, train_set, train_labels, n_max=None)\n",
    "e = time()\n",
    "print(\"Aux finished \" + str(e-s) + \" sec\")\n",
    "s = time()\n",
    "fill_set(G_s1, G_s2, train_set, train_labels, n_max=None)\n",
    "e = time()\n",
    "print(\"San finished \" + str(e-s) + \" sec\")\n",
    "pickle.dump( (train_set, train_labels), open( \"data/train_edge_face.set\", \"wb\" ) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_split=1e-07, min_samples_leaf=1,\n",
       "            min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "            n_estimators=400, n_jobs=8, oob_score=False, random_state=None,\n",
       "            verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=8)\n",
    "forest.fit(train_set, train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Start adding positive examples\n",
      "Finished : 0.09764309764309764\n",
      "Finished : 0.19528619528619529\n",
      "Finished : 0.29292929292929293\n",
      "Finished : 0.39057239057239057\n",
      "Finished : 0.4882154882154882\n",
      "Finished : 0.5858585858585859\n",
      "Finished : 0.6835016835016835\n",
      "Finished : 0.7811447811447811\n",
      "Finished : 0.8787878787878788\n",
      "Finished : 0.9764309764309764\n",
      "Start adding false examples\n",
      "Finished : 0.0\n",
      "Finished : 0.09829059829059829\n",
      "Finished : 0.19658119658119658\n",
      "Finished : 0.2948717948717949\n",
      "Finished : 0.39316239316239315\n",
      "Finished : 0.49145299145299143\n",
      "Finished : 0.5897435897435898\n",
      "Finished : 0.688034188034188\n",
      "Finished : 0.7863247863247863\n",
      "Finished : 0.8846153846153846\n",
      "Finished : 0.9829059829059829\n",
      "Test set generated in  0.5161821842193604 sec\n"
     ]
    }
   ],
   "source": [
    "test_set = []\n",
    "test_labels = []\n",
    "s = time()\n",
    "print(\"Dataset is generating ... \")\n",
    "fill_set(G_aux, G_san, test_set, test_labels, n_max=None)\n",
    "e = time()\n",
    "print(\"Test set generated in  \" + str(e-s) + \" sec\")\n",
    "pickle.dump( (test_set, test_labels), open( \"data/test_edge_face.set\", \"wb\" ) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.55932203389830504"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "forest.score(test_set, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "y_score = forest.predict_proba(test_set)\n",
    "y_true = test_labels\n",
    "fpr, tpr, thresholds = roc_curve(y_true, y_score[:, 1])\n",
    "roc_auc = auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiAAAAGHCAYAAACJeOnXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5f2b4ed9b0>"
      ]
     },
     "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": 81,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "y_score = forest.predict_proba(test_set)\n",
    "y_true = test_labels\n",
    "fpr, tpr, thresholds = roc_curve(y_true, y_score[:, 1])\n",
    "roc_auc = auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiAAAAGHCAYAAACJeOnXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\np7WjjZlnu70AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5f39827198>"
      ]
     },
     "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "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": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "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
}