-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathkerastest.py
215 lines (160 loc) · 10.3 KB
/
kerastest.py
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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
import numpy as np
import pandas as pd
import csv
from sklearn.model_selection import KFold, GridSearchCV
import time
from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay
import dask as dd
from sklearn.model_selection import train_test_split
from sklearn.metrics import f1_score
from sklearn.model_selection import cross_val_score
from sklearn.metrics import accuracy_score, mean_squared_error, classification_report, matthews_corrcoef
import sklearn
from sklearn import svm, ensemble
from sklearn.neighbors import KNeighborsClassifier
from sklearn.neural_network import MLPClassifier
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '1'
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
import pydot
from tensorflow.keras.wrappers.scikit_learn import KerasClassifier
from sklearn.model_selection import GridSearchCV
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
from tensorflow.keras.layers import Dropout
#predictors = [ensemble.RandomForestClassifier( random_state = 1, n_estimators = 100),ensemble.GradientBoostingClassifier(),
# KNeighborsClassifier(n_neighbors=10, n_jobs = 5, weights = 'distance' ), MLPClassifier()]
def get_rar_dataset(filename, n=None):
with open(filename) as data_file:
reader = csv.reader(data_file)
names = np.array(list(next(reader)))
data = pd.read_csv(filename, dtype=object)
data = data.to_numpy()
n = len(names) - 1
# ## Extract data names, membership values and Gram matrix
names = names[1:n+1]
mu = np.array([float(row[0]) for row in data[0:n+1]])
gram = np.array([[float(k.replace('NA', '0')) for k in row[1:n+1]]
for row in data[0:n+1]])
assert(len(names.shape) == 1)
assert(len(mu.shape) == 1)
assert(len(gram.shape) == 2)
assert(names.shape[0] == gram.shape[0] == gram.shape[1] == mu.shape[0])
X = np.array([[x] for x in np.arange(n)])
return X, gram, mu, names
file_name='mirrorcompareclassificationmatrixowlthing'
X, gram, mu, names = get_rar_dataset("/corese/axiom-prediction/fragments/"+file_name+".csv")
print('done extracting matrix')
# Function to create model, required for KerasClassifier
# def create_model(optimizer='Nadam',input_dim=0, init_mode='zero', activation='softmax',dropout_rate=0.8, neurons=20):
# model = Sequential()
# model.add(Dense(neurons, input_dim=input_dim, activation=activation))
# model.add(Dropout(dropout_rate))
# model.add(Dense(50, input_dim=neurons, activation=activation))
# #model.add(Dense(1,kernel_initializer=init_mode, activation='sigmoid'))
# model.add(Dense(1,kernel_initializer=init_mode, activation='sigmoid'))
# model.compile(loss='binary_crossentropy', optimizer=optimizer, metrics=['accuracy'])
# return model
def create_model(optimizer=keras.optimizers.RMSprop(),inputs = 0, init_mode='glorot_uniform', activation='relu', neurons=25):
model = Sequential()
model.add(Dense(neurons,input_shape=(inputs,),kernel_initializer=init_mode, activation=activation))
model.add(Dense(neurons*2,kernel_initializer=init_mode, activation=activation))
model.add(Dense(neurons/2,kernel_initializer=init_mode, activation=activation))
model.add(Dense(2))
#model.add(Dense(neurons/2,kernel_initializer=init_mode, activation=activation))
#keras.optimizers.RMSprop()
#model.add(Dense(1,kernel_initializer=init_mode, activation='sigmoid'))
model.compile(loss=keras.losses.SparseCategoricalCrossentropy(from_logits=True), optimizer=keras.optimizers.RMSprop(), metrics=['accuracy'])
#print(model.summary())
#keras.utils.plot_model(model, "my_first_model.png")
return model
print(file_name)
for i in range(3):
X_train, X_test, mu_train, mu_test = train_test_split(X, mu, test_size=0.9, stratify=mu)
train_test = gram[X_train.flatten()][:, X_train.flatten()]
test_test = gram[X_test.flatten()][:, X_train.flatten()]
test_names = names[X_test.flatten()]
inputs = keras.Input(shape=(len(X_train),))
model = KerasClassifier(build_fn=create_model,inputs = len(X_train),verbose=0, epochs= 100,batch_size=20 )
# define the grid search parameters
#batch_size = [10, 20, 40, 60, 80, 100]
#epochs = [10,20, 50, 100]
#optimizer = ['SGD', 'RMSprop', 'Adagrad', 'Adadelta', 'Adam', 'Adamax', 'Nadam']
#init_mode = ['uniform', 'lecun_uniform', 'normal', 'zero', 'glorot_normal', 'glorot_uniform', 'he_normal', 'he_uniform']
#activation = ['softmax', 'softplus', 'softsign', 'relu', 'tanh', 'sigmoid', 'hard_sigmoid', 'linear']
#dropout_rate = [0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9]
#neurons = [10, 15, 20, 25, 30]
#param_grid = dict(batch_size=batch_size, epochs=epochs, optimizer=optimizer, init_mode=init_mode,neurons=neurons, activation=activation, dropout_rate=dropout_rate)
# param_grid = dict(batch_size=batch_size,epochs = epochs)
# grid = GridSearchCV(estimator=model, param_grid=param_grid, cv=5, verbose = 3, n_jobs = 16)
# grid_result = grid.fit(train_test, mu_train)
history = model.fit(train_test, mu_train)
# summarize results
# print("Best: %f using %s" % (grid_result.best_score_, grid_result.best_params_))
# means = grid_result.cv_results_['mean_test_score']
# stds = grid_result.cv_results_['std_test_score']
# params = grid_result.cv_results_['params']
# for mean, stdev, param in zip(means, stds, params):
# print("%f (%f) with: %r" % (mean, stdev, param))
predicted_test_1 = model.predict(test_test)
print(classification_report(mu_test, predicted_test_1))
min_proba1 = model.predict_proba(test_test)
full_view_min = np.concatenate([np.vstack((test_names,mu_test, predicted_test_1.flatten())).T,min_proba1], axis = 1)
wrong_predictions_min = full_view_min[(full_view_min[:,1] != full_view_min[:,2])]
correct_predictions_min = full_view_min[(full_view_min[:,1] == full_view_min[:,2])]
ConfusionMatrixDisplay(confusion_matrix(mu_test, predicted_test_1),display_labels=['Rejected','Accepted']).plot()
# print(file_name)
# for i in range(1):
# X_train, X_test, mu_train, mu_test = train_test_split(X, mu, test_size=0.5, stratify=mu)
# train_test = gram[X_train.flatten()][:, X_train.flatten()]
# test_test = gram[X_test.flatten()][:, X_train.flatten()]
# test_names = names[X_test.flatten()]
# #crossval = cross_val_score(rs, gram[X.flatten()][:, X.flatten()], mu, cv=5)
# ticfirst = time.perf_counter()
# predictors[0].fit(train_test, mu_train)
# predictors[1].fit(train_test, mu_train)
# predictors[2].fit(train_test, mu_train)
# predictors[3].fit(train_test, mu_train)
# tocfirst = time.perf_counter()
# print(f"it took {tocfirst - ticfirst:0.4f} seconds")
# predicted_test_1 = predictors[0].predict(test_test)
# predicted_test_2 = predictors[1].predict(test_test)
# predicted_test_3 = predictors[2].predict(test_test)
# predicted_test_4 = predictors[3].predict(test_test)
# print("score: 1 " , matthews_corrcoef (mu_test, predicted_test_1))
# print("score: 2 " , matthews_corrcoef (mu_test, predicted_test_2))
# print("score: 3 " , matthews_corrcoef (mu_test, predicted_test_3))
# print("score: 4 " , matthews_corrcoef (mu_test, predicted_test_4))
# # predict_train= rs.predict(train_test)
# print(f'fold {i}:')
# print('test 1')
# print(classification_report(mu_test, predicted_test_1))
# print('test 2')
# print(classification_report(mu_test, predicted_test_2))
# print('test 3')
# print(classification_report(mu_test, predicted_test_3))
# print('test 4')
# print(classification_report(mu_test, predicted_test_4))
# # print('train')
# # print(classification_report(mu_train, predict_train))
# min_proba1 = predictors[0].predict_proba(test_test)
# min_proba2 = predictors[1].predict_proba(test_test)
# min_proba3 = predictors[2].predict_proba(test_test)
# min_proba4 = predictors[3].predict_proba(test_test)
# #print(crossval)
# #print("%0.2f accuracy with a standard deviation of %0.2f" % (crossval.mean(), crossval.std()))
# ConfusionMatrixDisplay(confusion_matrix(mu_test, predicted_test_1),display_labels=['Rejected','Accepted']).plot()
# ConfusionMatrixDisplay(confusion_matrix(mu_test, predicted_test_2),display_labels=['Rejected','Accepted']).plot()
# ConfusionMatrixDisplay(confusion_matrix(mu_test, predicted_test_3),display_labels=['Rejected','Accepted']).plot()
# ConfusionMatrixDisplay(confusion_matrix(mu_test, predicted_test_4),display_labels=['Rejected','Accepted']).plot()
# full_view_min = np.concatenate([np.vstack((test_names,mu_test, predicted_test_1,predicted_test_2,predicted_test_3,predicted_test_4)).T,min_proba1,min_proba2,min_proba3,min_proba4], axis = 1)
# wrong_predictions_min = full_view_min[(full_view_min[:,1] != full_view_min[:,2]) | (full_view_min[:,1] != full_view_min[:,3]) |(full_view_min[:,1] != full_view_min[:,4]) | (full_view_min[:,1] != full_view_min[:,5]) ]
# correct_predictions_min = full_view_min[(full_view_min[:,1] == full_view_min[:,2]) | (full_view_min[:,1] != full_view_min[:,3]) |(full_view_min[:,1] != full_view_min[:,4]) | (full_view_min[:,1] != full_view_min[:,5]) ]
# disagreement_wrong = wrong_predictions_min[(wrong_predictions_min[:,2] != wrong_predictions_min[:,3]) | (wrong_predictions_min[:,2] != wrong_predictions_min[:,4]) | (wrong_predictions_min[:,2] != wrong_predictions_min[:,5])
# | (wrong_predictions_min[:,3] != wrong_predictions_min[:,4]) | (wrong_predictions_min[:,3] != wrong_predictions_min[:,5]) |
# (wrong_predictions_min[:,4] != wrong_predictions_min[:,5])]
# disagreement_total = full_view_min[(full_view_min[:,2] != full_view_min[:,3]) | (full_view_min[:,2] != full_view_min[:,4]) | (full_view_min[:,2] != full_view_min[:,5])
# | (full_view_min[:,3] != full_view_min[:,4]) | (full_view_min[:,3] != full_view_min[:,5]) |
# (full_view_min[:,4] != full_view_min[:,5])]