ML-Deanon.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "%pylab inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "face_orig = open(\"data/facebook-links.txt\") "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "f_arr = np.array(face_orig.readlines())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "G = nx.Graph()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "edgelist = []\n",
    "for line in f_arr: \n",
    "    i, j, time = line.split('\\t')\n",
    "    edgelist.append((int(i), int(j)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "G.add_edges_from(edgelist)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "### Graph splitter "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def split_graph(G, alpha_v, save_as=\"\"):\n",
    "    V = np.array(G.nodes())\n",
    "    random.shuffle(V)\n",
    "    a = int((1-alpha_v)/2 * len(V))\n",
    "    b = int(alpha_v * len(V))\n",
    "    Va = V[:a]\n",
    "    Vb =  V[a:a+b]\n",
    "    Vc =  V[a+b:2*a+b]\n",
    "    V1 = np.concatenate((Va, Vb))\n",
    "    V2 = np.concatenate((Vb, Vc))\n",
    "    G1 = G.subgraph(V1)\n",
    "    G2 = G.subgraph(V2)\n",
    "    if save_as != '': \n",
    "        nx.write_edgelist(G1, 'data/'+str(alpha_v)+save_as+'_aux.edgelist')\n",
    "        nx.write_edgelist(G2, 'data/'+str(alpha_v)+save_as+'_san.edgelist')\n",
    "    return G1, G2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "G1, G2 = split_graph(G, 0.25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "## Apply anonimization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def anonimize(G):\n",
    "    return G"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "G_aux = anonimize(G1)\n",
    "G_san = anonimize(G2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generating training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "G_a1, G_a2 = split_graph(G_aux, 0.25) # aux \n",
    "G_s1, G_s2 = split_graph(G_san, 0.25) # sanitized"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "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",
    "def feature(G1, n, G2, m,  bins = 21, size = 50):\n",
    "    feature_set = [0 for i in range(4*bins)]\n",
    "    f_set = deg_dist(G1, n, bins, size)\n",
    "    for i in range(len(f_set)):\n",
    "        feature_set[i] += f_set[i]\n",
    "    f_set = deg_dist(G2, m, bins, size)\n",
    "    for i in range(len(f_set)):\n",
    "        feature_set[2*bins+i] += f_set[i]\n",
    "    # the Silhouette Coefficient \n",
    "    n_deg = G1.degree(n)\n",
    "    m_deg = G2.degree(m)\n",
    "    feature_set.append(abs(n_deg-m_deg)/max(n_deg, m_deg, 1))\n",
    "    return feature_set\n",
    "\n",
    "def fill_set(G1, G2, data, labels):\n",
    "    indx = 0\n",
    "    total = len(G1.nodes())\n",
    "    # report every 10 % \n",
    "    part = total//10\n",
    "    for i in G1.nodes():\n",
    "        if i in G2.nodes():\n",
    "            # There is an intersection\n",
    "            data.append(feature(G1, i, G2, i))\n",
    "            labels.append(1)\n",
    "        else:\n",
    "            # Choose randomly \n",
    "            j = random.choice(list(set(G2.nodes())-set([i])))\n",
    "            data.append(feature(G1, i, G2, j))\n",
    "            labels.append(0)\n",
    "        indx += 1\n",
    "        if indx % part ==0:\n",
    "            print(\"Finished : \" + str(indx/total))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Aux finished -252.8912250995636 sec\n",
      "San finished -284.2667410373688 sec\n"
     ]
    }
   ],
   "source": [
    "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",
    "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",
    "e = time()\n",
    "print(\"San finished \" + str(e-s) + \" sec\")\n",
    "pickle.dump( (train_set, train_labels), open( \"data/train_face.set\", \"wb\" ) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset is generating ... \n",
      "Finished : 0.09999748939268409\n",
      "Finished : 0.19999497878536818\n",
      "Finished : 0.29999246817805225\n",
      "Finished : 0.39998995757073635\n",
      "Finished : 0.49998744696342046\n",
      "Finished : 0.5999849363561045\n",
      "Finished : 0.6999824257487887\n",
      "Finished : 0.7999799151414727\n",
      "Finished : 0.8999774045341569\n",
      "Finished : 0.9999748939268409\n",
      "Test set generated in  814.7160470485687 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",
    "e = time()\n",
    "print(\"Test set generated in  \" + str(e-s) + \" sec\")\n",
    "pickle.dump( (test_set, test_labels), open( \"data/test_face.set\", \"wb\" ) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "forest = RandomForestClassifier(n_estimators = 400, n_jobs=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "collapsed": false
   },
   "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_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=400, n_jobs=4,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False)"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "forest.fit(train_set, train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "ada = RandomForestClassifier(n_estimators = 400, n_jobs=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "ada = AdaBoostClassifier(DecisionTreeClassifier(max_depth=3),\n",
    "                         n_estimators=400)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AdaBoostClassifier(algorithm='SAMME.R',\n",
       "          base_estimator=DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=3,\n",
       "            max_features=None, max_leaf_nodes=None, min_samples_leaf=1,\n",
       "            min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "            presort=False, random_state=None, splitter='best'),\n",
       "          learning_rate=1.0, n_estimators=400, random_state=None)"
      ]
     },
     "execution_count": 121,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ada.fit(train_set, train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.81820692425497732"
      ]
     },
     "execution_count": 122,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ada.score(test_set, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.82792297456754793"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "forest.score(test_set, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import GradientBoostingClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "gbc = GradientBoostingClassifier(n_estimators=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.82556300369059277"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gbc.fit(train_set, train_labels)\n",
    "gbc.score(test_set, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_set, train_labels = pickle.load(open( \"data/train_face.set\", \"rb\" ) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_set,  test_labels = pickle.load(open( \"data/test_face.set\", \"rb\" ) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Staring gridSearch ... \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=None, error_score='raise',\n",
       "       estimator=RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=10, n_jobs=-1,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False),\n",
       "       fit_params={}, iid=True, n_jobs=1,\n",
       "       param_grid={'max_features': ['auto'], 'criterion': ['gini'], 'n_estimators': [400], 'bootstrap': ['true']},\n",
       "       pre_dispatch='2*n_jobs', refit=True, scoring=None, verbose=0)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn import svm, datasets\n",
    "from sklearn.grid_search import GridSearchCV\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.cross_validation import cross_val_score as cv\n",
    "\n",
    "parameters = {'criterion':['gini'], 'n_estimators':[400], 'bootstrap':['true'], 'max_features':['auto']}\n",
    "svr = RandomForestClassifier(n_jobs=-1)\n",
    "print('Staring gridSearch ... ')\n",
    "clf = GridSearchCV(svr, parameters)\n",
    "clf.fit(train_set, train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap='true', class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features=None, max_leaf_nodes=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=600, n_jobs=-1,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.82767191383595695"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.score(test_set, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve, auc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "y_score = clf.predict(test_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "y_true = test_labels\n",
    "fpr, tpr, thresholds = roc_curve(y_true, y_score[:, 1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "roc_auc = auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiAAAAGHCAYAAACJeOnXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npGAtUTk3gQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1852c5198>"
      ]
     },
     "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": []
  }
 ],
 "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
}