ZprimeClassifier/PredictExternal.py at master · karavdin/ZprimeClassifier · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
import numpy as np
from numpy import inf
import keras
import matplotlib
import math
import matplotlib.pyplot as plt
from sklearn import preprocessing
from sklearn.metrics import roc_auc_score, roc_curve, auc
from sklearn.model_selection import train_test_split
from IPython.display import FileLink, FileLinks
from keras.models import Sequential
from keras.layers import Dense, Dropout, BatchNormalization
from keras.utils import to_categorical, plot_model
from keras.callbacks import History, ModelCheckpoint, ReduceLROnPlateau, EarlyStopping
from keras.optimizers import Adam
from keras import metrics, regularizers
import pickle
from copy import deepcopy
import os
from functions import *

def PredictExternal(parameters, inputfolder, outputfolder, filepostfix):
    print 'Making predictions now'
    tag = dict_to_str(parameters)
    classtag = get_classes_tag(parameters)
    fraction = get_fraction(parameters)

    # Get inputs
    model = keras.models.load_model(outputfolder+'/model.h5')
    model_best = keras.models.load_model(outputfolder+'/model_best.h5')
    input_train, input_test, input_val, labels_train, labels_test, labels_val, sample_weights_train, sample_weights_test, sample_weights_val, eventweights_train, eventweights_test, eventweights_val, signals, signal_eventweights, signal_normweights = load_data(parameters, inputfolder=inputfolder, filepostfix=filepostfix)

    signal_identifiers = ['RSGluon_All', 'RSGluon_M1000', 'RSGluon_M2000', 'RSGluon_M3000', 'RSGluon_M4000', 'RSGluon_M5000', 'RSGluon_M6000']

    # Do the predictions
    print 'Now that the model is trained, we\'re going to predict the labels of all 3 sets. '
    print 'predicting for training set'

    pred_train = model.predict(input_train)
    np.save(outputfolder+'/prediction_train.npy'  , pred_train)
    for cl in range(len(parameters['classes'])):
        print 'predicting for training set, class ' + str(cl)
        tmp = pred_train[labels_train[:,cl] == 1]
        np.save(outputfolder+'/prediction_train_class'+str(cl)+'.npy'  , tmp)
    print 'predicting for test set'
    print input_test.shape
    print labels_test.shape
    pred_test = model.predict(input_test)
    print pred_test.shape
    print labels_test.shape
    np.save(outputfolder+'/prediction_test.npy'  , pred_test)
    for cl in range(len(parameters['classes'])):
        print 'predicting for test set, class ' + str(cl)
        tmp = pred_test[labels_test[:,cl] == 1]
        np.save(outputfolder+'/prediction_test_class'+str(cl)+'.npy'  , tmp)
    print 'predicting for val set'
    pred_val = model.predict(input_val)
    np.save(outputfolder+'/prediction_val.npy'  , pred_val)
    for cl in range(len(parameters['classes'])):
        print 'predicting for val set, class ' + str(cl)
        tmp = pred_val[labels_val[:,cl] == 1]
        np.save(outputfolder+'/prediction_val_class'+str(cl)+'.npy'  , tmp)

    # Do predictions with best model instead of last
    print 'predicting for training set, best model'
    pred_train = model_best.predict(input_train)
    np.save(outputfolder+'/prediction_train_best.npy'  , pred_train)
    for cl in range(len(parameters['classes'])):
        print 'predicting for training set, best model, class ' + str(cl)
        tmp = pred_train[labels_train[:,cl] == 1]
        np.save(outputfolder+'/prediction_train_class'+str(cl)+'_best.npy'  , tmp)
    print 'predicting for test set, best model'
    pred_test = model_best.predict(input_test)
    np.save(outputfolder+'/prediction_test_best.npy'  , pred_test)
    for cl in range(len(parameters['classes'])):
        print 'predicting for test set, best model, class ' + str(cl)
        tmp = pred_test[labels_test[:,cl] == 1]
        np.save(outputfolder+'/prediction_test_class'+str(cl)+'_best.npy'  , tmp)
    print 'predicting for val set, best model'
    pred_val = model_best.predict(input_val)
    np.save(outputfolder+'/prediction_val_best.npy'  , pred_val)
    for cl in range(len(parameters['classes'])):
        print 'predicting for val set, best model, class ' + str(cl)
        tmp = pred_val[labels_val[:,cl] == 1]
        np.save(outputfolder+'/prediction_val_class'+str(cl)+'_best.npy'  , tmp)

    print 'predicting for signals'
    for i in range(len(signal_identifiers)):
        pred_signal= model.predict(signals[i])
        np.save(outputfolder+'/prediction_'+signal_identifiers[i]+'.npy'  , pred_signal)
        pred_signal = model_best.predict(signals[i])
        np.save(outputfolder+'/prediction_'+signal_identifiers[i]+'_best.npy'  , pred_signal)


def PredictExternalOnPredictions(parameters, inputfolder, inputfolder_predictions, outputfolder, filepostfix):
    print 'Making predictions now'
    tag = dict_to_str(parameters)
    classtag = get_classes_tag(parameters)
    fraction = get_fraction(parameters)

    # Get inputs
    model = keras.models.load_model(outputfolder+'/model.h5')
    model_best = keras.models.load_model(outputfolder+'/model_best.h5')
    input_train, input_test, input_val, labels_train, labels_test, labels_val, sample_weights_train, sample_weights_test, sample_weights_val, eventweights_train, eventweights_test, eventweights_val, signals, signal_eventweights, signal_normweights = load_data(parameters, inputfolder=inputfolder, filepostfix='')

    input_train, input_test, input_val, signals = load_predictions(outputfolder=inputfolder_predictions, filepostfix=filepostfix)
    # pred_train, pred_test, pred_val, pred_signals = load_predictions(outputfolder=inputfolder_predictions, filepostfix=filepostfix)
    # input_train = np.concatenate((input_train, pred_train), axis=1)
    # input_test = np.concatenate((input_test, pred_test), axis=1)
    # input_val = np.concatenate((input_val, pred_val), axis=1)
    # for i in signals.keys():
    #     signals[i] = np.concatenate((signals[i], pred_signals[i]), axis=1)

    signal_identifiers = ['RSGluon_All', 'RSGluon_M1000', 'RSGluon_M2000', 'RSGluon_M3000', 'RSGluon_M4000', 'RSGluon_M5000', 'RSGluon_M6000']

    # Do the predictions
    print 'Now that the model is trained, we\'re going to predict the labels of all 3 sets. '
    print 'predicting for training set'

    pred_train = model.predict(input_train)
    np.save(outputfolder+'/prediction_train.npy'  , pred_train)
    for cl in range(len(parameters['classes'])):
        print 'predicting for training set, class ' + str(cl)
        tmp = pred_train[labels_train[:,cl] == 1]
        np.save(outputfolder+'/prediction_train_class'+str(cl)+'.npy'  , tmp)
    print 'predicting for test set'
    pred_test = model.predict(input_test)
    np.save(outputfolder+'/prediction_test.npy'  , pred_test)
    for cl in range(len(parameters['classes'])):
        print 'predicting for test set, class ' + str(cl)
        tmp = pred_test[labels_test[:,cl] == 1]
        np.save(outputfolder+'/prediction_test_class'+str(cl)+'.npy'  , tmp)
    print 'predicting for val set'
    pred_val = model.predict(input_val)
    np.save(outputfolder+'/prediction_val.npy'  , pred_val)
    for cl in range(len(parameters['classes'])):
        print 'predicting for val set, class ' + str(cl)
        tmp = pred_val[labels_val[:,cl] == 1]
        np.save(outputfolder+'/prediction_val_class'+str(cl)+'.npy'  , tmp)

    # Do predictions with best model instead of last
    print 'predicting for training set, best model'
    pred_train = model_best.predict(input_train)
    np.save(outputfolder+'/prediction_train_best.npy'  , pred_train)
    for cl in range(len(parameters['classes'])):
        print 'predicting for training set, best model, class ' + str(cl)
        tmp = pred_train[labels_train[:,cl] == 1]
        np.save(outputfolder+'/prediction_train_class'+str(cl)+'_best.npy'  , tmp)
    print 'predicting for test set, best model'
    pred_test = model_best.predict(input_test)
    np.save(outputfolder+'/prediction_test_best.npy'  , pred_test)
    for cl in range(len(parameters['classes'])):
        print 'predicting for test set, best model, class ' + str(cl)
        tmp = pred_test[labels_test[:,cl] == 1]
        np.save(outputfolder+'/prediction_test_class'+str(cl)+'_best.npy'  , tmp)
    print 'predicting for val set, best model'
    pred_val = model_best.predict(input_val)
    np.save(outputfolder+'/prediction_val_best.npy'  , pred_val)
    for cl in range(len(parameters['classes'])):
        print 'predicting for val set, best model, class ' + str(cl)
        tmp = pred_val[labels_val[:,cl] == 1]
        np.save(outputfolder+'/prediction_val_class'+str(cl)+'_best.npy'  , tmp)

    print 'predicting for signals'
    for i in range(len(signal_identifiers)):
        pred_signal= model.predict(signals[i])
        np.save(outputfolder+'/prediction_'+signal_identifiers[i]+'.npy'  , pred_signal)
        pred_signal = model_best.predict(signals[i])
        np.save(outputfolder+'/prediction_'+signal_identifiers[i]+'_best.npy'  , pred_signal)