Hhh_optimization_Apply_regression.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# KERAS and TENSORFLOW-THEANO\n",
    "TUTORIAL: Applying the regression previously obtained\n",
    "(%lsmagic for magic commanda and ! for bash instruction)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using Theano backend.\n"
     ]
    },
    {
     "data": {
      "application/javascript": [
       "\n",
       "require(['notebook'],\n",
       "  function() {\n",
       "    IPython.CodeCell.config_defaults.highlight_modes['magic_text/x-c++src'] = {'reg':[/^%%cpp/]};\n",
       "    console.log(\"JupyROOT - %%cpp magic configured\");\n",
       "  }\n",
       ");\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Welcome to JupyROOT 6.08/06\n",
      "If you had no error so far, this is great! We can start the tutorial.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/Luca2/Library/Python/2.7/lib/python/site-packages/root_numpy/__init__.py:34: RuntimeWarning: ROOT 6.08/06 is currently active but you installed root_numpy against ROOT 6.08/00. Please consider reinstalling root_numpy for this ROOT version.\n",
      "  RuntimeWarning)\n",
      "/Users/Luca2/Library/Python/2.7/lib/python/site-packages/root_numpy/__init__.py:46: RuntimeWarning: numpy 1.12.1 is currently installed but you installed root_numpy against numpy 1.11.0. Please consider reinstalling root_numpy for this numpy version.\n",
      "  RuntimeWarning)\n"
     ]
    }
   ],
   "source": [
    "# We start by importing all the necessary packages\n",
    "import sys;\n",
    "import os, sys, array, re, math, random, subprocess, glob\n",
    "from math import *\n",
    "import numpy as np\n",
    "import scipy\n",
    "from numpy.lib.recfunctions import stack_arrays\n",
    "import pandas as pd\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import cPickle\n",
    "from sklearn.feature_selection import SelectKBest, f_classif\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV\n",
    "from sklearn.preprocessing import StandardScaler, RobustScaler\n",
    "from sklearn.utils import compute_class_weight\n",
    "from sklearn.model_selection import KFold\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Dense, Activation, Dropout, Highway, MaxoutDense, Masking, GRU, Merge, Input, merge\n",
    "from keras.callbacks import EarlyStopping, ModelCheckpoint\n",
    "from keras.utils.np_utils import to_categorical\n",
    "from keras.wrappers.scikit_learn import KerasRegressor\n",
    "import deepdish.io as io\n",
    "import ROOT\n",
    "from ROOT import gSystem, gROOT, gApplication, TFile, TTree, TCut, TH1F, TCanvas\n",
    "from root_numpy import root2array \n",
    "from IPython.display import HTML, IFrame\n",
    "import seaborn as sns; sns.set()\n",
    "print \"I you had no error so far, this is great! We can start the tutorial.\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We set some variables we will use later"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('', None)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check this file out, it contains many functions we will use.\n",
    "execfile(\"Useful_func.py\")\n",
    "# We start loading the regression weights\n",
    "model_Lead    = load_model('models/Weights_regression_LeadingJet.h5')\n",
    "model_SubLead = load_model('models/Weights_regression_SubleadingJet.h5')\n",
    "# Parameters\n",
    "folder='Plots_Regression_Applied/' # Folder with Plots\n",
    "MakePlots=True # Set False if you want to run faster\n",
    "folderCreation  = subprocess.Popen(['mkdir -p ' + folder], stdout=subprocess.PIPE, shell=True); folderCreation.communicate()\n",
    "folderCreation2 = subprocess.Popen(['mkdir -p models/'], stdout=subprocess.PIPE, shell=True); folderCreation2.communicate()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now We Start manipulating ROOT files into a format we can use to train a MVA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---> hh Displayed as panda dataframe: \n",
      "      numOfVertices    b1jet_pt  b1jet_eta    b1jet_mt  b1jet_leadTrackPt  \\\n",
      "0                19  173.124298  -0.714916  173.738297          47.906250   \n",
      "1                33  246.682297   0.298446  248.075989          58.843750   \n",
      "2                30  340.403229   1.258016  342.380005          98.562500   \n",
      "3                14   80.552818   2.200674   81.073486          22.250000   \n",
      "4                16   86.618324  -1.492352   87.986191          10.437500   \n",
      "5                29  147.646530   0.765988  148.593613          38.468750   \n",
      "6                20  179.774414  -0.576904  181.731323          33.406250   \n",
      "7                13  160.337036  -0.464420  162.175751          36.437500   \n",
      "8                16   67.824074   1.615699   68.589928          27.625000   \n",
      "9                13  118.275604  -1.762509  119.216217          32.031250   \n",
      "10               42  174.260849   1.073892  176.100327          26.484375   \n",
      "11               33  142.049393   2.365291  142.999451          32.187500   \n",
      "12               12  124.097084  -1.880355  124.704475          38.593750   \n",
      "13               12  143.997528   0.601616  144.939636          24.234375   \n",
      "14               18   71.668770  -0.621273   73.713371          11.593750   \n",
      "15               17  168.491440  -0.539859  171.719299          32.968750   \n",
      "16               19   73.634766   0.321159   74.633331          17.312500   \n",
      "17               21  119.909294  -1.196250  120.385368          49.750000   \n",
      "18               23   61.311272   1.524936   62.131676          10.976562   \n",
      "19               16   61.002487   0.815358   62.718700           9.234375   \n",
      "20               19  139.507935  -1.512107  141.882187          27.843750   \n",
      "21               28  110.306679  -0.054426  111.114868          26.453125   \n",
      "22               24  104.096039   1.401750  105.602226          18.250000   \n",
      "23               25  291.652802   0.259874  292.595581         127.875000   \n",
      "24               19   81.992393   1.780108   82.384354          52.343750   \n",
      "25               30  142.617386  -2.139359  145.040131          16.468750   \n",
      "26               23  116.806374   1.657782  117.575821          19.390625   \n",
      "27               27  154.985306  -0.641958  155.757401          39.375000   \n",
      "28               13  165.338593  -0.083306  166.437698          73.812500   \n",
      "29               13   92.013832  -1.531184   93.055374          18.718750   \n",
      "...             ...         ...        ...         ...                ...   \n",
      "4179             10  106.920563   1.595005  107.600555          33.718750   \n",
      "4180             14  168.977829   0.369030  169.942062          20.859375   \n",
      "4181             17  162.309174  -0.738977  163.393280          50.562500   \n",
      "4182             18  195.396683   0.986165  197.111237          51.656250   \n",
      "4183             15  145.113007   0.869500  145.971710          42.187500   \n",
      "4184             21  132.413589  -0.916839  133.690887          23.671875   \n",
      "4185              9  156.645050  -0.551639  157.811218          30.000000   \n",
      "4186             14   64.592789   0.469875   65.037506          26.578125   \n",
      "4187             13   91.033875  -1.366691   91.870544          13.710938   \n",
      "4188             19  208.564087  -0.688823  209.781982          39.437500   \n",
      "4189             13  131.006027  -0.635232  131.669830          38.250000   \n",
      "4190             19  290.065521   0.494386  294.082397          70.250000   \n",
      "4191             14  138.806778  -1.244819  139.321594          42.781250   \n",
      "4192             17   73.083168  -0.811610   73.370872          18.265625   \n",
      "4193             11   89.882904   0.667456   90.569321          19.734375   \n",
      "4194             23  146.358154  -1.029058  147.400726          36.937500   \n",
      "4195             15   92.057373  -0.101632   92.881844          28.609375   \n",
      "4196             17  104.442467  -0.505933  105.075981          24.078125   \n",
      "4197             22   97.425247  -1.337259   98.079651          36.718750   \n",
      "4198             17   53.973141   0.510759   54.886993           6.984375   \n",
      "4199             22  252.435120   0.119302  253.330368          49.406250   \n",
      "4200             21   34.874130   1.170465   35.407772           8.390625   \n",
      "4201             15   72.169090   1.631212   72.741409          10.757812   \n",
      "4202             16  112.309776   2.326932  114.263672          20.156250   \n",
      "4203             18  215.564163  -1.329550  216.930481          38.218750   \n",
      "4204             10  183.669724  -0.545932  184.282684          32.468750   \n",
      "4205             24  142.690216  -0.111881  145.287308          18.406250   \n",
      "4206             24  194.084244  -1.529882  195.278931          42.531250   \n",
      "4207             20   58.787407   1.038611   59.727310          12.554688   \n",
      "4208             24  205.237549   0.429516  206.807175          67.312500   \n",
      "\n",
      "      b1jet_leptonDeltaR  b1jet_leptonPtRel  b1jet_leptonPt  b1jet_vtxPt  \\\n",
      "0               8.886246           0.000000       -1.000000    74.793633   \n",
      "1               9.475537           0.000000       -1.000000     0.000000   \n",
      "2              10.648453           0.000000       -1.000000    32.354179   \n",
      "3               0.030109           0.616346       20.328125     0.000000   \n",
      "4               8.461505           0.000000       -1.000000     0.000000   \n",
      "5              10.325735           0.000000       -1.000000    51.358002   \n",
      "6               8.449391           0.000000       -1.000000    59.352242   \n",
      "7               8.993304           0.000000       -1.000000    39.031902   \n",
      "8              11.567269           0.000000       -1.000000     9.160653   \n",
      "9               0.240124           1.074758        4.089844    36.400349   \n",
      "10             11.049647           0.000000       -1.000000    37.085175   \n",
      "11              0.345360           0.914004        2.976562    74.579536   \n",
      "12              7.274754           0.000000       -1.000000    74.031311   \n",
      "13             10.610615           0.000000       -1.000000    46.212391   \n",
      "14              0.192709           2.019463       10.992188    24.498207   \n",
      "15              0.196314           5.027470       24.781250    31.539532   \n",
      "16              0.145149           1.424389        9.976562    21.788664   \n",
      "17              8.662917           0.000000       -1.000000    26.808952   \n",
      "18             11.824152           0.000000       -1.000000    22.394354   \n",
      "19              0.208740           1.845764        9.234375    16.530825   \n",
      "20              0.080805           0.395285        4.757812    33.233883   \n",
      "21              9.439346           0.000000       -1.000000    39.979744   \n",
      "22              0.065805           0.926004       14.359375    22.372431   \n",
      "23              9.902719           0.000000       -1.000000    72.094505   \n",
      "24              0.022394           1.172604       52.343750     0.000000   \n",
      "25              8.440980           0.000000       -1.000000    48.842617   \n",
      "26              0.024754           0.316189       12.851562    28.930849   \n",
      "27              8.358043           0.000000       -1.000000   112.758125   \n",
      "28              0.130784           0.720070        5.515625    10.892419   \n",
      "29              7.470647           0.000000       -1.000000     0.000000   \n",
      "...                  ...                ...             ...          ...   \n",
      "4179            0.028555           0.975030       33.718750    13.896862   \n",
      "4180           10.113488           0.000000       -1.000000    35.457756   \n",
      "4181            8.778753           0.000000       -1.000000    29.944033   \n",
      "4182           10.057064           0.000000       -1.000000     0.000000   \n",
      "4183           10.051836           0.000000       -1.000000    67.497658   \n",
      "4184            8.448045           0.000000       -1.000000    18.176243   \n",
      "4185            0.018033           0.542842       30.000000    32.532211   \n",
      "4186            0.008087           0.214524       26.578125     0.000000   \n",
      "4187            0.045076           0.336845        7.582031    44.351162   \n",
      "4188            8.950663           0.000000       -1.000000     0.000000   \n",
      "4189            9.048704           0.000000       -1.000000    55.797474   \n",
      "4190           10.442945           0.000000       -1.000000    58.515644   \n",
      "4191            0.079511           0.883417       11.085938    80.296387   \n",
      "4192            8.420338           0.000000       -1.000000    31.049994   \n",
      "4193            9.693789           0.000000       -1.000000    26.811209   \n",
      "4194            9.134704           0.000000       -1.000000    62.402737   \n",
      "4195            9.511589           0.000000       -1.000000     0.000000   \n",
      "4196            8.527472           0.000000       -1.000000    63.183777   \n",
      "4197            9.638962           0.000000       -1.000000    49.043892   \n",
      "4198            9.727963           0.000000       -1.000000    26.755766   \n",
      "4199            9.206501           0.000000       -1.000000   158.867340   \n",
      "4200            0.074485           0.643987        8.390625    13.533996   \n",
      "4201            0.012889           0.137996       10.757812    39.250118   \n",
      "4202           11.417752           0.000000       -1.000000     0.000000   \n",
      "4203            0.143067           0.792131        5.882812    55.851654   \n",
      "4204            9.968075           0.000000       -1.000000    62.366951   \n",
      "4205            8.903337           0.000000       -1.000000    24.979242   \n",
      "4206            7.471152           0.000000       -1.000000    21.232380   \n",
      "4207           10.287676           0.000000       -1.000000    37.401752   \n",
      "4208            9.606818           0.000000       -1.000000    88.483246   \n",
      "\n",
      "      b1jet_vtxMass  b1jet_vtxNtracks  b1jet_neHEF  b1jet_neEmEF  \\\n",
      "0          1.621181               5.0     0.050416      0.484596   \n",
      "1          0.000000               0.0     0.138334      0.352227   \n",
      "2          0.974425               2.0     0.008021      0.458380   \n",
      "3          0.000000               0.0     0.098938      0.361311   \n",
      "4          0.000000               0.0     0.081685      0.191647   \n",
      "5          1.497231               4.0     0.000000      0.416437   \n",
      "6          1.739681               5.0     0.032811      0.386302   \n",
      "7          2.327658               7.0     0.063804      0.437253   \n",
      "8          1.250745               4.0     0.014668      0.188212   \n",
      "9          1.130635               2.0     0.052190      0.266156   \n",
      "10         0.912831               3.0     0.057075      0.405371   \n",
      "11         2.521918               6.0     0.036332      0.330625   \n",
      "12         2.496731               5.0     0.109807      0.068393   \n",
      "13         1.682473               4.0     0.056433      0.471207   \n",
      "14         3.971229               4.0     0.068335      0.153292   \n",
      "15         1.394474               2.0     0.092126      0.104681   \n",
      "16         1.927003               3.0     0.125832      0.456496   \n",
      "17         1.102442               2.0     0.466162      0.208322   \n",
      "18         0.993611               4.0     0.140550      0.309787   \n",
      "19         2.311434               3.0     0.017308      0.294034   \n",
      "20         0.787441               2.0     0.074661      0.264439   \n",
      "21         1.805104               3.0     0.114604      0.424745   \n",
      "22         0.865642               2.0     0.173671      0.134720   \n",
      "23         3.475206               6.0     0.025497      0.533626   \n",
      "24         0.000000               0.0     0.026820      0.085760   \n",
      "25         2.747471               8.0     0.060475      0.125024   \n",
      "26         1.147392               2.0     0.068261      0.046258   \n",
      "27         3.618078               9.0     0.035768      0.170569   \n",
      "28         0.922683               3.0     0.042418      0.633090   \n",
      "29         0.000000               0.0     0.019250      0.427386   \n",
      "...             ...               ...          ...           ...   \n",
      "4179       0.787464               2.0     0.000000      0.063075   \n",
      "4180       1.646550               4.0     0.000000      0.084273   \n",
      "4181       2.206645               4.0     0.021163      0.497783   \n",
      "4182       0.000000               0.0     0.058190      0.487911   \n",
      "4183       2.562554               4.0     0.035697      0.176890   \n",
      "4184       0.754159               3.0     0.195004      0.270035   \n",
      "4185       0.721098               3.0     0.198016      0.294027   \n",
      "4186       0.000000               0.0     0.145080      0.176597   \n",
      "4187       2.464221               4.0     0.011300      0.222361   \n",
      "4188       0.000000               0.0     0.064089      0.355883   \n",
      "4189       3.461878               7.0     0.000000      0.363505   \n",
      "4190       1.418736               4.0     0.270883      0.353508   \n",
      "4191       2.654532               3.0     0.009667      0.244694   \n",
      "4192       1.733792               4.0     0.058730      0.194406   \n",
      "4193       2.271355               4.0     0.140116      0.445143   \n",
      "4194       2.253494               3.0     0.050462      0.122176   \n",
      "4195       0.000000               0.0     0.072155      0.356835   \n",
      "4196       2.896321               4.0     0.054089      0.115826   \n",
      "4197       3.103243               4.0     0.148426      0.204084   \n",
      "4198       3.157123               7.0     0.127100      0.228040   \n",
      "4199       2.972229               5.0     0.036805      0.272501   \n",
      "4200       1.520187               4.0     0.078941      0.100557   \n",
      "4201       2.054717               5.0     0.007880      0.166400   \n",
      "4202       0.000000               0.0     0.000000      0.455082   \n",
      "4203       1.523466               3.0     0.079329      0.192642   \n",
      "4204       2.829094               6.0     0.000000      0.185033   \n",
      "4205       2.729821               4.0     0.051243      0.215653   \n",
      "4206       0.376182               2.0     0.177323      0.331266   \n",
      "4207       4.190404               6.0     0.093170      0.147334   \n",
      "4208       2.671451               3.0     0.079875      0.287074   \n",
      "\n",
      "      b1jet_vtx3DSig  b1jet_vtx3DVal  b1genjet_pt    target  \n",
      "0          26.157988        1.967698   172.178650  0.994538  \n",
      "1           0.000000        0.000000   195.914536  0.794198  \n",
      "2           7.705590        0.944695   304.522736  0.894594  \n",
      "3           0.000000        0.000000    54.073444  0.671279  \n",
      "4           0.000000        0.000000    81.036469  0.935558  \n",
      "5          10.634332        0.517075   160.419449  1.086510  \n",
      "6           7.832742        0.336888   190.524002  1.059795  \n",
      "7         111.891029        1.903939   145.027847  0.904519  \n",
      "8          15.613366        1.055085    40.859528  0.602434  \n",
      "9           4.221067        0.729125    29.835064  0.252250  \n",
      "10         59.300240        3.875120    25.439190  0.145983  \n",
      "11          9.813640        2.269392    54.086449  0.380758  \n",
      "12         12.895526        1.508751   148.629974  1.197691  \n",
      "13         70.620102        2.019354   142.146790  0.987147  \n",
      "14          6.774051        0.119047    77.364563  1.079474  \n",
      "15         36.233181        1.567695    38.814510  0.230365  \n",
      "16         34.668743        0.558064    75.007530  1.018643  \n",
      "17          4.029860        0.370166   112.457726  0.937857  \n",
      "18         12.429051        1.608558   103.063805  1.680993  \n",
      "19         14.351771        0.375071    26.262325  0.430512  \n",
      "20         35.490208        3.891939   160.244614  1.148642  \n",
      "21         25.943197        0.521567    96.823601  0.877767  \n",
      "22          7.299679        0.716665   196.157196  1.884387  \n",
      "23          6.141870        0.121861    58.760139  0.201473  \n",
      "24          0.000000        0.000000    94.045120  1.146998  \n",
      "25         95.135391        9.070968   140.353271  0.984125  \n",
      "26         11.310946        1.180041   100.526291  0.860623  \n",
      "27         95.211067        2.471222   146.179962  0.943186  \n",
      "28         29.623516        0.763060    34.241428  0.207099  \n",
      "29          0.000000        0.000000    66.349541  0.721082  \n",
      "...              ...             ...          ...       ...  \n",
      "4179       80.747337        5.091310    97.303169  0.910051  \n",
      "4180       12.049495        0.340086    63.751541  0.377278  \n",
      "4181       65.610970        1.475026    18.061647  0.111279  \n",
      "4182        0.000000        0.000000   168.360474  0.861634  \n",
      "4183      137.901337        3.142104    55.107250  0.379754  \n",
      "4184       28.632401        3.441884    98.093506  0.740811  \n",
      "4185       20.371883        2.223397    72.392227  0.462142  \n",
      "4186        0.000000        0.000000    43.591057  0.674860  \n",
      "4187        9.091996        0.358959   108.653152  1.193546  \n",
      "4188        0.000000        0.000000    36.969624  0.177258  \n",
      "4189       32.269066        0.708640    78.951546  0.602656  \n",
      "4190       62.424313        2.307542    76.908180  0.265141  \n",
      "4191       48.107204        2.382466    41.935692  0.302116  \n",
      "4192       28.372293        0.905632    51.823368  0.709101  \n",
      "4193      105.291374        1.804788    76.075172  0.846381  \n",
      "4194       86.123398        3.343389   145.684204  0.995395  \n",
      "4195        0.000000        0.000000    72.891304  0.791803  \n",
      "4196       24.550695        0.510523    94.095932  0.900936  \n",
      "4197       86.534584        2.708078    99.048744  1.016664  \n",
      "4198       24.108347        0.302249    43.435471  0.804761  \n",
      "4199       31.803873        1.445006    40.416618  0.160107  \n",
      "4200        9.507304        0.457221    34.610790  0.992449  \n",
      "4201       39.083225        2.107584    84.251884  1.167423  \n",
      "4202        0.000000        0.000000    38.743317  0.344968  \n",
      "4203       83.166000        4.362824    69.819923  0.323894  \n",
      "4204       17.599173        0.385763    73.018883  0.397555  \n",
      "4205       12.401098        0.151244    87.586975  0.613826  \n",
      "4206        6.663843        4.151495    23.140509  0.119229  \n",
      "4207       32.566494        0.527393    52.341438  0.890351  \n",
      "4208       20.452982        0.582686    75.035706  0.365604  \n",
      "\n",
      "[4209 rows x 17 columns]\n",
      "The shape for LeadJet is (samples, features): \n",
      "(4209, 17)\n",
      "The shape for SubLeadJet is (samples, features): \n",
      "(4209, 17)\n"
     ]
    }
   ],
   "source": [
    "# 0) We first have to place in alist the files we would like to apply the regression too.\n",
    "Input_Files = ['MVA_GluGluToRadionToHHTo2B2VTo2L2Nu_M-500_narrow_13TeV-madgraph-v2.root']\n",
    "# 1) Place a selection if you want to use it\n",
    "my_selec = 'met_pt>20 && met_pt<500 && muon1_pt>20 && muon2_pt>10 && fabs(muon1_eta)<2.4 && fabs(muon2_eta)<2.4 && pt_l1l2<500 && pt_b1b2<500 && mass_l1l2>12 && mass_l1l2<500 && mass_b1b2<500 && b1jet_pt>20 && fabs(b1jet_eta)<2.4 && b2jet_pt>20 && fabs(b2jet_eta)<2.4 && mass_trans>10 && mass_trans<500 && HT<4000'\n",
    "# 2) Selecting the branches that contains the information we want to use (in general)\n",
    "my_branches_jet1 = [\"numOfVertices\",\"b1jet_pt\",\"b1jet_eta\",\"b1jet_mt\",\"b1jet_leadTrackPt\",\"b1jet_leptonDeltaR\",\"b1jet_leptonPtRel\",\"b1jet_leptonPt\",\"b1jet_vtxPt\",\"b1jet_vtxMass\",\"b1jet_vtxNtracks\",\"b1jet_neHEF\",\"b1jet_neEmEF\",\"b1jet_vtx3DSig\",\"b1jet_vtx3DVal\",\"b1genjet_pt\"]\n",
    "my_branches_jet2 = [\"numOfVertices\",\"b2jet_pt\",\"b2jet_eta\",\"b2jet_mt\",\"b2jet_leadTrackPt\",\"b2jet_leptonDeltaR\",\"b2jet_leptonPtRel\",\"b2jet_leptonPt\",\"b2jet_vtxPt\",\"b2jet_vtxMass\",\"b2jet_vtxNtracks\",\"b2jet_neHEF\",\"b2jet_neEmEF\",\"b2jet_vtx3DSig\",\"b2jet_vtx3DVal\",\"b2genjet_pt\"]\n",
    "# 3) Selecting the branches that contains the information we want to use (in the training)\n",
    "my_branches_training_jet1 = my_branches_jet1\n",
    "my_branches_training_jet2 = my_branches_jet2\n",
    "# Converting Root files in dataframe (Very useful, checnl root2panda in Useful_func.py)\n",
    "LeadJet    = root2panda('files'+Input_Files[0], 'DiHiggsWWBBAna/evtree', branches=my_branches_jet1, selection=my_selec_jet1)\n",
    "SubLeadJet = root2panda('files'+Input_Files[0], 'DiHiggsWWBBAna/evtree', branches=my_branches_jet2, selection=my_selec_jet2)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You need:\n",
    "1. X : matrix with raw=#Events and column=Variables to discriminate.\"\n",
    "2. w : A vector containig the weights of each event\"\n",
    "3. Y : A vector containing the target: genPt/recoPt. This is the weight you will then apply to the recoPt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Now lets start to talk about DNN!\n"
     ]
    }
   ],
   "source": [
    "print('Now lets start to talk about DNN!')\n",
    "# Turn the df the desired ndarray \"X\" that can be directly used for ML applications.\n",
    "X    = LeadJet[my_branches_training_jet1].as_matrix() # Each row is an object to classify, each column corresponds to a specific variable.\n",
    "if (doSubLead):\n",
    "    X = SubLeadJet[my_branches_training_jet2].as_matrix()\n",
    "# No weights needed, just set al to 1\n",
    "w    =  np.ones(X.shape[0])\n",
    "# This is the array with the true values: 0 is signal, 1 if TT.\n",
    "y = LeadJet[\"target\"]\n",
    "if (doSubLead):\n",
    "  y = SubLeadJet[\"target\"]\n",
    "\n",
    "# Randomly shuffle and automatically split all your objects into train and test subsets\n",
    "ix = range(X.shape[0]) # array of indices, just to keep track of them for safety reasons and future checks\n",
    "X_train, X_test, y_train, y_test, w_train, w_test, ix_train, ix_test = train_test_split(X, y, w, ix, train_size=0.7) # Train here is 70% of the total statistic\n",
    "# It is common practice to scale the inputs to Neural Nets such that they have approximately similar ranges (it atually improve the results)\n",
    "scaler = StandardScaler() \n",
    "X_train = scaler.fit_transform(X_train)\n",
    "X_test = scaler.transform(X_test) # You are applying the same transformation done to X_train, to X_test."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "IOPub data rate exceeded.\n",
      "The notebook server will temporarily stop sending output\n",
      "to the client in order to avoid crashing it.\n",
      "To change this limit, set the config variable\n",
      "`--NotebookApp.iopub_data_rate_limit`.\n"
     ]
    }
   ],
   "source": [
    "# This takes a while, but it is worth do do it once. It shows the correlations for S and B overimposed.\n",
    "# More variables you add in var_toPlot, more times it takes.\n",
    "if MakePlots:\n",
    "    # Next lines to select the same number of entries\n",
    "    var_toPlot = [\"target\",\"numOfVertices\",\"b1jet_pt\",\"b1jet_eta\",\"b1jet_mt\",\"b1jet_leadTrackPt\",\"b1jet_leptonDeltaR\",\"b1jet_leptonPtRel\",\"b1jet_leptonPt\",\"b1jet_vtxPt\",\"b1jet_vtxMass\",\"b1jet_vtxNtracks\",\"b1jet_neHEF\",\"b1jet_neEmEF\",\"b1jet_vtx3DSig\",\"b1jet_vtx3DVal\",\"b1genjet_pt\"]\n",
    "    if (doSubLead):\n",
    "        var_toPlot = [\"target\",\"numOfVertices\",\"b2jet_pt\",\"b2jet_eta\",\"b2jet_mt\",\"b2jet_leadTrackPt\",\"b2jet_leptonDeltaR\",\"b2jet_leptonPtRel\",\"b2jet_leptonPt\",\"b2jet_vtxPt\",\"b2jet_vtxMass\",\"b2jet_vtxNtracks\",\"b2jet_neHEF\",\"b2jet_neEmEF\",\"b2jet_vtx3DSig\",\"b2jet_vtx3DVal\",\"b2genjet_pt\"]\n",
    "    # You can select the variable to plot in sns.pairplot using an argumnet vars=['var1','var2'...]\n",
    "    sns_plot = sns.pairplot(LeadJet,palette=[\"#e74c3c\"],plot_kws={\"s\": 3,\"alpha\":0.3},size=5)\n",
    "    sns_plot.savefig(folder + \"/Variables_pairplot_LeadingJet.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x1575d0e10>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Now fitting the Train sample to get our model\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 10)                170       \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 10)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 20)                220       \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 20)                0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 20)                420       \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 20)                0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 10)                210       \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 10)                0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 20)                220       \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 20)                0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 1)                 21        \n",
      "=================================================================\n",
      "Total params: 1,261\n",
      "Trainable params: 1,261\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Regression\n",
    "def baseline_model():\n",
    "    model = Sequential()\n",
    "    model.add(Dense(10, input_dim=X_train.shape[1], activation='relu')) # Linear transformation of the input vector. The first number is output_dim.\\n\",\n",
    "    model.add(Dropout(0.1)) # To avoid overfitting. It masks the outputs of the previous layer such that some of them will randomly become inactive and will not contribute to infor\n",
    "    model.add(Dense(20, activation='relu'))\n",
    "    model.add(Dropout(0.1))\n",
    "    model.add(Dense(20, activation='sigmoid'))\n",
    "    model.add(Dropout(0.1))\n",
    "    model.add(Dense(10,activation='tanh'))\n",
    "    model.add(Dropout(0.1))\n",
    "    model.add(Dense(20, activation='sigmoid'))\n",
    "    model.add(Dropout(0.1))\n",
    "    model.add(Dense(1))\n",
    "    # Compile model\n",
    "    model.compile(loss='mean_squared_error', optimizer='adam')\n",
    "    model.summary()\n",
    "    return model\n",
    "# evaluate model with standardized dataset\n",
    "estimator = KerasRegressor(build_fn=baseline_model, nb_epoch=100, batch_size=15, verbose=0)\n",
    "print \"Now fitting the Train sample to get our model\"\n",
    "estimator.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA8MAAAKvCAYAAABONB2jAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x157845b10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA84AAAKvCAYAAAC7ypYTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEz4yMucAAAADSURBVE2/1GNr9IsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x14423e090>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA84AAAKvCAYAAAC7ypYTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkiRJkpowcJYkSZIkqYn/AaQo0e3QQt/RAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1450560d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score: 0.1970435606\n",
      "('Saving:', 'Plots_Regression//Regression_y_hat.pdf')\n",
      "('Saving:', 'Plots_Regression//Regression_y_hat_OverY.pdf')\n",
      "WARNING. Pt Feature not found!!\n"
     ]
    }
   ],
   "source": [
    "# Get a prediction on the test\n",
    "y_hat = estimator.predict(X_test)\n",
    "# Get the score\n",
    "score = estimator.score(X_test, y_test)\n",
    "print \"score:\", score\n",
    "\n",
    "# Plot the test target and the estimated one\n",
    "%matplotlib inline\n",
    "if MakePlots:\n",
    "    # Estimatd target\n",
    "    matplotlib.rcParams.update({'font.size': 16})\n",
    "    fig = plt.figure(figsize=(11.69, 8.27), dpi=100)\n",
    "    bins = np.linspace(my_max(min(y_hat),0.), max(y_hat), 100)\n",
    "    _ = plt.hist(y_hat,  bins=bins, histtype='step', normed=True, label=r'$y_hat$', linewidth=2)\n",
    "    plt.xlabel(\"y_hat\")\n",
    "    plt.ylabel('Entries')\n",
    "    plt.legend(loc='best')\n",
    "    print('Saving:',folder + '/Regression_y_hat.pdf')\n",
    "    plt.savefig(folder + '/Regression_y_hat.pdf')\n",
    "    # Estimatd target/ True target\n",
    "    matplotlib.rcParams.update({'font.size': 16})\n",
    "    fig = plt.figure(figsize=(11.69, 8.27), dpi=100)\n",
    "    bins = np.linspace(my_max(min(y_hat/y_test),0.), 2, 100)\n",
    "    _ = plt.hist(y_hat/y_test,  bins=bins, histtype='step', normed=True, label=r'$y_hat/y$', linewidth=2)\n",
    "    plt.xlabel(\"y_hat/y\")\n",
    "    plt.ylabel('Entries')\n",
    "    plt.legend(loc='best')\n",
    "    print('Saving:',folder + '/Regression_y_hat_OverY.pdf')\n",
    "    plt.savefig(folder + '/Regression_y_hat_OverY.pdf')\n",
    "    # Pt Resolution Before After (first I have to find the index in X_test that corresponf to the Jet and Genjet pT)\n",
    "    N_bjet_pt = -1; N_bgenjet_pt = -1\n",
    "    N_tmp = 0\n",
    "    for Feature in my_branches_training_jet1:\n",
    "        if(not doSubLead and Feature==\"b1jet_pt\"): N_bjet_pt = N_tmp\n",
    "        if(not doSubLead and Feature==\"b1genjet_pt\"): N_bgenjet_pt = N_tmp\n",
    "        if(doSubLead and Feature==\"b2jet_pt\"): N_bjet_pt = N_tmp\n",
    "        if(doSubLead and Feature==\"b2genjet_pt\"): N_bgenjet_pt = N_tmp\n",
    "        N_tmp += 1\n",
    "    if( N_bjet_pt==-1 or N_bgenjet_pt==-1 ): print \"WARNING. Pt Feature not found!!\"\n",
    "    X_test_ori = scaler.inverse_transform(X_test) # Need also to trasform back the X_test\n",
    "    bjet_pt = X_test_ori[:,N_bjet_pt]\n",
    "    bgenjet_pt = X_test_ori[:,N_bgenjet_pt]\n",
    "    Variable       = (bjet_pt-bgenjet_pt)       \n",
    "    Variable_corr  = (bjet_pt*y_hat-bgenjet_pt) \n",
    "    Variable_ideal = (bjet_pt*y_test-bgenjet_pt)\n",
    "    matplotlib.rcParams.update({'font.size': 16})\n",
    "    fig = plt.figure(figsize=(11.69, 8.27), dpi=100)\n",
    "    bins = np.linspace(-50, 50, 100)\n",
    "    _ = plt.hist(Variable,  bins=bins, histtype='step', normed=True, label=r'$STD$', linewidth=2)\n",
    "    _ = plt.hist(Variable_corr,  bins=bins, histtype='step', normed=True, label=r'$CORR$', linewidth=2)\n",
    "    #You chan check if this below is a single Bin in 0\n",
    "    #_ = plt.hist(Variable_ideal,  bins=bins, histtype='step', normed=True, label=r'$IDEAL$', linewidth=2)\n",
    "    plt.xlabel(\"pT Resolution\")\n",
    "    plt.ylabel('Entries')\n",
    "    plt.legend(loc='best')\n",
    "    plt.savefig(folder + '/Regression_PtRes.pdf')\n",
    "\n",
    "#kfold = KFold(n_splits=10, random_state=seed)\n",
    "#results = cross_val_score(estimator, X, Y, cv=kfold)\n",
    "#print(\"Results: %.2f (%.2f) MSE\" % (results.mean(), results.std()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'model' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mNameError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-10-73680bf1e8f5>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mname\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m\"Weights_regression_LeadingJet.h5\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mdoSubLead\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mname\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m\"Weights_regression_SubleadingJet.h5\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 4\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msave\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfolder\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;34m'/'\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'model' is not defined"
     ]
    }
   ],
   "source": [
    "# Now save this model into a file\n",
    "name = \"Weights_regression_LeadingJet.h5\"\n",
    "if doSubLead: name = \"Weights_regression_SubleadingJet.h5\"\n",
    "model.save(folder + '/' + name)"
   ]
  }
 ],
 "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": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}