Analyze count feat, minpt, TMVA.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:2659a36de18757d18d1507b0cc2a7ad3ab9b7b313d48f53498b4d8a4450c0e9e"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# About\n",
      "\n",
      "Fix some parameters for Matrixnet and AdaBoost. Fix 0.2 size for signal training (instead 0.5). Bck and signal test is 0.5 from all dataset  \n",
      "\n",
      "\n",
      "1) Compare 7 features for training and all features\n",
      "\n",
      "\n",
      "2) Compare train on dataset and on dataset with svr.minpt > 500\n",
      "\n",
      "\n",
      "3) Test 2) on \n",
      "\n",
      "\n",
      "    a) all test dataset \n",
      "    b) for svr.minpt <= 500 we say that this not pass presel \n",
      "    \n",
      "4) Compare TMVA and Sklearn AdaBoost"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "channels_all = [\"mod_11114001.csv\", \"mod_11124001.csv\", \n",
      "                \"mod_11296013.csv\", \"mod_11874042.csv\", \n",
      "                \"mod_12103035.csv\", \"mod_12165106.csv\", \n",
      "                \"mod_12265042.csv\", \"mod_30000000.csv\"]\n",
      "folder = \"Topo/csv_data/\""
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import io\n",
      "from IPython.nbformat import current\n",
      "\n",
      "def execute_notebook(nbfile):\n",
      "    \n",
      "    with io.open(nbfile) as f:\n",
      "        nb = current.read(f, 'json')\n",
      "    \n",
      "    ip = get_ipython()\n",
      "    \n",
      "    for cell in nb.worksheets[0].cells:\n",
      "        if cell.cell_type != 'code':\n",
      "            continue\n",
      "        ip.run_cell(cell.input)\n",
      "\n",
      "        \n",
      "import sys\n",
      "sys.path.insert(0, '/local/userdata/shelf/notebook/antares/work/REP')\n",
      "import cern_utils\n",
      "from cern_utils import converter_csv\n",
      "\n",
      "execute_notebook('metric_generate.ipynb')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# head = True\n",
      "# for channel in channels_all:\n",
      "#     t = pandas.read_csv(folder + channel, sep='\\t')\n",
      "#     t.to_csv(folder + 'signal.csv', header=head, index=False, sep='\\t', mode='a')\n",
      "#     head = False"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Variables"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "variables = [\"sumpt\",\n",
      "             \"m\",\n",
      "             \"mcor\",\n",
      "             \"ipchi2\",\n",
      "             \"vchi2\",\n",
      "             \"sumipchi2\",\n",
      "             \"fdr\",\n",
      "             \"nlt16\",\n",
      "             \"minpt\",\n",
      "             \"ptau\",\n",
      "             \"q\",\n",
      "             \"eta\",\n",
      "             \"nhlt1\",\n",
      "             \"nmu\",\n",
      "             \"n\",\n",
      "             \"fdchi2\"]\n",
      "\n",
      "variables_small = [\"sumpt\",\n",
      "             \"m\",\n",
      "             \"mcor\",\n",
      "             \"ipchi2\",\n",
      "             \"vchi2\",\n",
      "             \"fdchi2\",\n",
      "             \"minpt\"]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 16
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "sig_data = converter_csv.load_from_csv(folder + 'signal.csv', sep='\\t')\n",
      "bck_data = converter_csv.load_from_csv(folder + channels_all[-1], sep='\\t')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 6
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def get_channel_labels(ds):\n",
      "    channel_label = []\n",
      "\n",
      "    for element in ds['# event_id']:\n",
      "        channel_label.append(int(element.split('_')[0]))\n",
      "    return channel_label"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 7
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.preprocessing import LabelEncoder\n",
      "from sklearn import preprocessing\n",
      "le = preprocessing.LabelEncoder()\n",
      "le.fit(map(lambda x: int(x[4:-4]), channels_all))\n",
      "sig_data = sig_data.add_columns({'channel': le.transform(get_channel_labels(sig_data)), \n",
      "                                 'channel_label': get_channel_labels(sig_data)})\n",
      "bck_data = bck_data.add_columns({'channel': le.transform(get_channel_labels(bck_data)),\n",
      "                                 'channel_label': get_channel_labels(bck_data)})"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 8
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "bck_data_train, bck_data_test = bck_data.get_train_test(train_size=0.5, \n",
      "                                                        test_size=0.5, group_columns='# event_id')\n",
      "sig_data_train, sig_data_test = sig_data.get_train_test(train_size=0.2, test_size=0.5, group_columns='# event_id')\n",
      "bck_data_train = bck_data_train.select('presel==1')\n",
      "sig_data_train = sig_data_train.select('presel==1')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 9
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "len(sig_data), len(bck_data), len(sig_data_train), len(bck_data_train)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 12,
       "text": [
        "(11011932, 1562996, 667524, 201421)"
       ]
      }
     ],
     "prompt_number": 12
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Matrixnet training"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import sklearn\n",
      "from cern_utils import sklearn_classifier, ef_classifier"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 13
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_feat = ef_classifier.ClassifierEF(directory='train_anal/mn_feat_all', formula_name='topo_mn_train', \n",
      "                                     dataset_name='topo_mn_train_all_features', url_base='test02cern.vs.os.yandex.net')\n",
      "mn_feat.set_params(features=variables, iterations=5000, sync=False)\n",
      "mn_feat.fit(sig_data_train, bck_data_train)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 14,
       "text": [
        "<cern_utils.ef_classifier.ClassifierEF at 0x10dd1b10>"
       ]
      }
     ],
     "prompt_number": 14
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_feat7 = ef_classifier.ClassifierEF(directory='train_anal/mn_feat_7', formula_name='topo_mn_train', \n",
      "                                     dataset_name='topo_mn_train_7_features', url_base='test02cern.vs.os.yandex.net')\n",
      "mn_feat7.set_params(features=variables_small, iterations=5000, sync=False)\n",
      "mn_feat7.fit(sig_data_train, bck_data_train)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 17,
       "text": [
        "<cern_utils.ef_classifier.ClassifierEF at 0x13211890>"
       ]
      }
     ],
     "prompt_number": 17
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_minpt = ef_classifier.ClassifierEF(directory='train_anal/mn_minpt', formula_name='topo_mn_train', \n",
      "                                     dataset_name='topo_mn_train_minpt', url_base='test02cern.vs.os.yandex.net')\n",
      "mn_minpt.set_params(features=variables, iterations=5000, sync=False)\n",
      "mn_minpt.fit(sig_data_train.select('minpt > 500'), bck_data_train.select('minpt > 500'))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 19,
       "text": [
        "<cern_utils.ef_classifier.ClassifierEF at 0x5f95150>"
       ]
      }
     ],
     "prompt_number": 19
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Ada Boost training"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.ensemble import AdaBoostClassifier\n",
      "from sklearn.tree import DecisionTreeClassifier"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 24
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ada_feat = sklearn_classifier.ClassifierSklearn(base_classifier=AdaBoostClassifier(n_estimators=400, \n",
      "                                                                                   learning_rate=0.01, random_state=13,\n",
      "                              base_estimator=DecisionTreeClassifier(max_depth=6, min_samples_leaf=30, max_features=6,\n",
      "                                                                    random_state=13)),\n",
      "                                                directory='train_anal/ada_feat_all')\n",
      "ada_feat.set_params(features=variables)\n",
      "ada_feat.fit(sig_data_train, bck_data_train)\n",
      "\n",
      "ada_feat7 = sklearn_classifier.ClassifierSklearn(base_classifier=AdaBoostClassifier(n_estimators=400, \n",
      "                                                                                   learning_rate=0.01, random_state=13,\n",
      "                              base_estimator=DecisionTreeClassifier(max_depth=6, min_samples_leaf=30, max_features=6,\n",
      "                                                                    random_state=13)),\n",
      "                                                  directory='train_anal/ada_feat_7')\n",
      "ada_feat7.set_params(features=variables_small)\n",
      "ada_feat7.fit(sig_data_train, bck_data_train)\n",
      "\n",
      "ada_minpt = sklearn_classifier.ClassifierSklearn(base_classifier=AdaBoostClassifier(n_estimators=400, \n",
      "                                                                                   learning_rate=0.01, random_state=13,\n",
      "                              base_estimator=DecisionTreeClassifier(max_depth=6, min_samples_leaf=30, max_features=6,\n",
      "                                                                    random_state=13)),\n",
      "                                                  directory='train_anal/ada_minpt')\n",
      "ada_minpt.set_params(features=variables)\n",
      "ada_minpt.fit(sig_data_train.select('minpt > 500'), bck_data_train.select('minpt > 500'))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "test_all = sig_data_test.union([sig_data_test, bck_data_test])\n",
      "eff_to_plot = metric_generate(test_all.get_data(), metric_plot=True)\n",
      "signal_labels = [1] * len(sig_data_test) + [0] * len(bck_data_test)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "182\n"
       ]
      }
     ],
     "prompt_number": 29
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "test_all_mod = test_all.get_data()\n",
      "presel = numpy.array(test_all_mod['presel'])\n",
      "presel[numpy.array(test_all_mod['minpt']) <= 500] = 0\n",
      "test_all_mod['presel'] = presel\n",
      "eff_to_plot_mod = metric_generate(test_all_mod, metric_plot=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "182\n"
       ]
      }
     ],
     "prompt_number": 42
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mn_probs = {}\n",
      "mn_probs['all features'] = mn_feat.predict(test_all)\n",
      "mn_probs['7 features'] = mn_feat7.predict(test_all)\n",
      "mn_probs['minpt > 500 for train'] = mn_minpt.predict(test_all)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 31
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ada_probs = {}\n",
      "ada_probs['all features'] = ada_feat.predict(test_all)\n",
      "ada_probs['7 features'] = ada_feat7.predict(test_all)\n",
      "ada_probs['minpt > 500 for train'] = ada_minpt.predict(test_all)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 32
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Compare channels"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "to_plot_res = {}\n",
      "num = 0\n",
      "color = ['b', 'g', 'cyan', 'pink', 'orange', 'black', 'yellow', 'brown']\n",
      "for key, prob_mn in mn_probs.iteritems():\n",
      "    to_plot_res[key] = {'data': eff_to_plot(signal_labels, prob_mn), \n",
      "                                  'color': color[num], 'move': 3 + num, 'alpha': 1,\n",
      "                                  'p_label': 'p_empty', 'total': 'total_empty'}\n",
      "    num += 1"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "182\n",
        "0.977143238508\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.976113801962\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.977225631756\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.0\n",
        "0\n"
       ]
      }
     ],
     "prompt_number": 36
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "to_plot_res['BDT']['move'] = 8"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 45
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "to_plot_res['minpt > 500 for train (cut on test minpt > 500)'] = {'data': eff_to_plot_mod(signal_labels, \n",
      "                                                                                       mn_probs['minpt > 500 for train']), \n",
      "                                  'color': color[4], 'move': 3 + 3, 'alpha': 1,\n",
      "                                  'p_label': 'p_empty', 'total': 'total_empty'}\n",
      "to_plot_res['all features (cut on test minpt > 500)'] = {'data': eff_to_plot_mod(signal_labels, \n",
      "                                                                                       mn_probs['all features']), \n",
      "                                  'color': color[5], 'move': 3 + 4, 'alpha': 1,\n",
      "                                  'p_label': 'p_empty', 'total': 'total_empty'}\n",
      "to_plot_res['BDT'] = {'data': eff_to_plot(signal_labels, numpy.zeros(len(signal_labels))), 'color': 'r', 'move': 3 + 5,\n",
      "                      'p_label': 'bdt_passed', 'total': 'total_empty'}"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "182\n",
        "0.974525007783\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.974931501703\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.0\n",
        "0\n"
       ]
      }
     ],
     "prompt_number": 47
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plt.subplots(figsize=(13,7))\n",
      "plot_eff(to_plot_res)\n",
      "plt.xlabel('Modes')\n",
      "plt.ylabel('Inefficiency')\n",
      "plt.title('Classifier Inefficiency for Matrixnet')\n",
      "plt.legend(loc='best')\n",
      "plt.tight_layout()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA58AAAHyCAYAAABs5HrRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x202dcc10>"
       ]
      }
     ],
     "prompt_number": 48
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "to_plot_res_ada = {}\n",
      "num = 0\n",
      "color = ['b', 'g', 'cyan', 'pink', 'orange', 'black', 'yellow', 'brown']\n",
      "for key, prob_ada in ada_probs.iteritems():\n",
      "    to_plot_res_ada[key] = {'data': eff_to_plot(signal_labels, prob_ada), \n",
      "                                  'color': color[num], 'move': 3 + num, 'alpha': 1,\n",
      "                                  'p_label': 'p_empty', 'total': 'total_empty'}\n",
      "    num += 1\n",
      "    \n",
      "to_plot_res_ada['BDT'] = to_plot_res['BDT']"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "182\n",
        "0.709692255271\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.70778589586\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.70296480302\n",
        "182\n"
       ]
      }
     ],
     "prompt_number": 39
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "to_plot_res_ada['minpt > 500 for train (cut on test minpt > 500)'] = {'data': eff_to_plot_mod(signal_labels, \n",
      "                                                                                       ada_probs['minpt > 500 for train']), \n",
      "                                  'color': color[4], 'move': 3 + 3, 'alpha': 1,\n",
      "                                  'p_label': 'p_empty', 'total': 'total_empty'}\n",
      "to_plot_res_ada['all features (cut on test minpt > 500)'] = {'data': eff_to_plot_mod(signal_labels, \n",
      "                                                                                       ada_probs['all features']), \n",
      "                                  'color': color[5], 'move': 3 + 4, 'alpha': 1,\n",
      "                                  'p_label': 'p_empty', 'total': 'total_empty'}\n",
      "to_plot_res_ada['BDT'] = to_plot_res['BDT']"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "182\n",
        "0.70159267281\n",
        "182\n",
        "182"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "0.697039566128\n",
        "182\n"
       ]
      }
     ],
     "prompt_number": 49
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plt.subplots(figsize=(13,7))\n",
      "plot_eff(to_plot_res_ada)\n",
      "plt.xlabel('Modes')\n",
      "plt.ylabel('Inefficiency')\n",
      "plt.title('Classifier Inefficiency for AdaBoost')\n",
      "plt.legend(loc='best')\n",
      "plt.tight_layout()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA58AAAHyCAYAAABs5HrRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0xeb30ed0>"
       ]
      }
     ],
     "prompt_number": 50
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import cPickle\n",
      "with open('train_anal/ada_compare', 'w') as file:\n",
      "    cPickle.dump(to_plot_res_ada, file)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 58
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import cPickle\n",
      "with open('train_anal/mn_compare', 'w') as file:\n",
      "    cPickle.dump(to_plot_res, file)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 59
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Compare TMVA Ada and Sklearn Ada\n",
      "\n",
      "If take Ada TMVA with the same learining rate as in Sklearn (0.01) then we can't find threshold, which gives us 2,5 rate on bck. Threshold was equal 1., so inefficincy = 1. That is why I put learning rate 0.04, so this works better, but Ada in Sklearn is better.\n",
      "\n",
      "__In fact maybe I did't do complete mapping of sklearn parameters to tmva. Please, check it if you can!__"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## TMVA - \n",
      "\n",
      "`\"!H:!V:NTrees=400:BoostType=AdaBoost:MaxDepth=6:AdaBoostBeta=0.04:nCuts=20:NEventsMin=30\"`"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## Sklearn - \n",
      "\n",
      "`n_estimators=400, learning_rate=0.01, random_state=13, base_estimator=DecisionTreeClassifier(max_depth=6, min_samples_leaf=30, max_features=6, random_state=13)`"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from cern_utils import tmva_classifier"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 53
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tmva = tmva_classifier.ClassifierTMVA('train_anal/tmva_ada')\n",
      "tmva.set_params(tmva_method = ('ADA', \n",
      "                                \"!H:!V:NTrees=400:BoostType=AdaBoost:MaxDepth=6:AdaBoostBeta=0.01:nCuts=20:NEventsMin=30\", \n",
      "                                'kBDT'),\n",
      "                features=variables)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tmva.fit(sig_data_train, bck_data_train)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tmva2 = tmva_classifier.ClassifierTMVA('train_anal/tmva_ada2')\n",
      "tmva2.set_params(tmva_method = ('ADA', \n",
      "                                \"!H:!V:NTrees=400:BoostType=AdaBoost:MaxDepth=6:AdaBoostBeta=0.04:nCuts=20:NEventsMin=30\", \n",
      "                                'kBDT'),\n",
      "                features=variables)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 95
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tmva2.fit(sig_data_train, bck_data_train)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 96,
       "text": [
        "<cern_utils.tmva_classifier.ClassifierTMVA at 0x7f78ff626b10>"
       ]
      }
     ],
     "prompt_number": 96
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tmva_prob = tmva.predict(test_all)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 64
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tmva_prob2 = tmva2.predict(test_all)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 97
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "compare_ada_tmva = {}\n",
      "compare_ada_tmva['ada sklearn'] = to_plot_res_ada['all features']\n",
      "compare_ada_tmva['ada tmva'] = {'data': eff_to_plot(signal_labels, tmva_prob2), \n",
      "                                  'color': color[0], 'move': 3 + 3, 'alpha': 1,\n",
      "                                  'p_label': 'p_empty', 'total': 'total_empty'}"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "182\n",
        "0.992777059892\n",
        "182\n"
       ]
      }
     ],
     "prompt_number": 98
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plt.subplots(figsize=(13,7))\n",
      "plot_eff(compare_ada_tmva)\n",
      "plt.xlabel('Modes')\n",
      "plt.ylabel('Inefficiency')\n",
      "plt.title('Classifier Inefficiency for AdaBoost')\n",
      "plt.legend(loc='best')\n",
      "plt.tight_layout()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA6EAAAHyCAYAAAAa3shaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x1320a090>"
       ]
      }
     ],
     "prompt_number": 99
    }
   ],
   "metadata": {}
  }
 ]
}