-
Notifications
You must be signed in to change notification settings - Fork 0
/
neuralnetworksubmissionmaker.py
316 lines (253 loc) · 9.52 KB
/
neuralnetworksubmissionmaker.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
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
import numpy as np
import pathlib
from datetime import datetime
# classifiers
from sklearn.ensemble import AdaBoostClassifier, GradientBoostingClassifier, ExtraTreesClassifier, RandomForestClassifier
from sklearn.linear_model import LogisticRegression, PassiveAggressiveClassifier, RidgeClassifier, SGDClassifier
from sklearn.linear_model import LogisticRegressionCV
from sklearn.neighbors import KNeighborsClassifier
from sklearn.neural_network import BernoulliRBM, MLPClassifier
from sklearn.naive_bayes import GaussianNB
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.gaussian_process import GaussianProcessClassifier
from sklearn.gaussian_process.kernels import RBF
from sklearn.cross_validation import KFold
from sklearn.model_selection import GridSearchCV, ShuffleSplit
from sklearn.preprocessing import Normalizer
from sklearn.pipeline import Pipeline
# import matplotlib.pyplot as plt
def load_data(filename, train=True):
"""
Function loads data stored in the file filename and returns it as a numpy ndarray.
Inputs:
filename: given as a string
(optional) train: used to determine whether this is the training or test set
Outputs:
Data contained in the file, returned as a numpy ndarray
"""
X = []
y = []
with open(filename) as f:
for line in f:
if (train):
# remove \n, split on space, separate into label and weights
X.append(line.strip().split(' ')[1:])
y.append(line.strip().split(' ')[0])
else:
X.append(line.strip().split(' '))
# convert to np, cast to int, and remove the headers
X = np.asarray(X[1:]).astype(int)
if (train):
y = np.asarray(y[1:]).astype(int)
return X, y
def split_data(x_train, y_train):
'''
Function for cross validiation.
Inputs:
x_train: training data points
y_train: training labels
Outputs:
trainX: randomized 4/5 of given data points
trainY: corresponding labels
testX: randomized 1/5 of given data points
testY: corresponding lables
'''
dataSplit = ShuffleSplit(n_splits = 1, test_size = 0.2)
for train, test in dataSplit.split(x_train, y_train):
return x_train[train], y_train[train], x_train[test], y_train[test]
def normalization(X_train, X_test):
'''
Function to normalize training and test data
Inputs:
X_train: training set data points
X_test: test set data points
Outputs:
train_norm: normalized training set data points
test_norm: normalized test set data points
'''
normalizer = Normalizer().fit(X_train)
train_norm = normalizer.transform(X_train)
test_norm = normalizer.transform(X_test)
return (train_norm, test_norm)
def make_predictions(clf, X, y, test):
'''
Function to train and test our classifier
Inputs:
clf: classifier
X: data points
y: labels
test: test set
Outputs:
predictions: predictions from running the clf on the test set
'''
clf.fit(X, y)
predictions = clf.predict(test)
predictions = predictions.astype(int)
return predictions
def save_data(data, filename="%s.txt" % datetime.today().strftime("%X").replace(":", "")):
'''
Function to save the predictions by the classifier
Inputs: predictions, (optional) filename
If filename isn't specified, then it just uses the current time
Outputs: Does not return anything
Writes the submisssion to a textfile that should have the same format as the sample_submission.txt
'''
# Creates a new submissions folder if one doesn't exist
pathlib.Path('submissions').mkdir(parents=True, exist_ok=True)
with open("submissions\\%s" % filename, "w") as f:
f.write("Id,Prediction\n")
for Id, prediction in enumerate(data, 1):
string = str(Id) + ',' + str(prediction) + '\n'
f.write(string)
def percentError(yPred, yTrue):
'''
Calculates the percent error between two given label sets
Inputs:
yPred: predicted labels
yTrue: actual labels
Outputs:
error: float of the number of mismatches divided by total length
'''
return 1.0-np.sum(np.equal(yPred, yTrue))/len(yTrue)
from keras.models import Sequential
from keras.layers import Dense, Activation, Dropout
from keras.layers import Conv2D, MaxPooling2D, Flatten, BatchNormalization
from keras.layers.advanced_activations import LeakyReLU, PReLU
from keras import regularizers
from keras.utils import to_categorical
def main():
# load the data
X_train, y_train = load_data("training_data.txt")
X_test, _ = load_data("test_data.txt", False)
# normalize training and test data
X_train_n, X_test_n = normalization(X_train, X_test)
# split the data in to training and testing so we can test ourselves
trainX, trainY, testX, testY = split_data(X_train_n, y_train)
y_binary = to_categorical(y_train)
rate = 0.5
model = Sequential()
model.add(Dense(300, input_shape=(1000,), activation='sigmoid'))
# model.add(Activation('sigmoid'))
# model.add(BatchNormalization())
model.add(Dropout(rate))
model.add(Dense(300, activation='relu',
kernel_regularizer=regularizers.l2(0.001)
# activity_regularizer=regularizers.l2(0.005)
))
# model.add(LeakyReLU(alpha=.01))
# model.add(Dropout(rate))
# output layer
model.add(Dense(2, activation='softmax'))
## Printing a summary of the layers and weights in your model
model.summary()
model.compile(loss='binary_crossentropy', optimizer='Adam', metrics=['accuracy'])
fit = model.fit(X_train_n, y_binary, batch_size=32, epochs=50, verbose=1)
# testY_binary = to_categorical(testY)
## Printing the accuracy of our model, according to the loss function specified in model.compile above
# score = model.evaluate(testX, testY_binary, verbose=0)
# print('Test score:', score[0])
# print('Test accuracy:', score[1])
seq_binarypredictions = model.predict(X_test_n)
save_data(seq_binarypredictions, "neuralbinarypreds.txt")
seq_predictions = []
for i in seq_binarypredictions:
if i[0] > i[1]:
seq_predictions.append(0)
elif i[0] < i[1]:
seq_predictions.append(1)
else:
print(i[0], i[1], "are equal!!!!\n\n")
save_data(seq_predictions, "NeuralNetworkSubmission.txt")
if __name__ == '__main__':
main()
'''
kernelreg l2=0.001
160, 160, 3 .852, 0.8297
160, 160, 10 .85475
160, 160, 30, 64 .84325
kernelreg l2=0.01
160, 160, 5 .8525, 0.85425
160, 160, 6 .855, 0.844
kernelreg l1=0.001
160, 160, 3 0.85325, 0.84075b, 0.8d
160, 160, 5 0.84375, 0.85025
160, 160, 10 .8515, 0.854
160, 160, 15 0.84375
160, 160, 20 .85075
activityreg l2=0.0001
160, 160, 15 .84525
d = 0
150, 4, b .846, .85125
150, 6, b .854, .84975, .8465
200, 8 .847
200, 10 .84825
200, 12 .85175
d = 0.3
150, 4, d .8565, .8535, .85625
150, 6, d .849
150, 8, d .852, .83875
150, 10, d .8575
dropout = 0.5
150, 10, d .85, 853, .84975
150, 10, d, separate softmax .84225, .84725
200, 2, no d .84975, .84575
200, 3, no d .8385, .851
200, 4, no d .8495, .85225
200, 8, d .85575, .855, .85, .8585, .84675
200, 8, no d .847
200, 10, d .847, .8505, .85425
200, 10, no d .84825
200, 10, d, sep .8525
200, 12, no d .85175
130, 130, 10 .8465, .8505
130, 130, 12 .84625
130, 130, 15 .862, .851
130, 150, 10 .852, .8465, .84975
130, 150, 12 .85025, .85475, .8455
150, 130, 8 .854
150, 130, 10 .82925, .852, .8425, .85175
150, 130, 11 .86425, .85325, .85625, .84575, .8435, .84625, .8515, 0.8525, .849
150, 130, 12 .85425, 0.86275, .8505, 0.85925, .8425, 0.85125
150, 130, 13 .852, 0.84475
150, 130, 15 .8455
150, 150, 10, 64 .83825, 0.851
150, 150, 13, 16 .851, 0.8565
150, 150, 14, 64 .849, 0.85475
150, 150, 14, 128 .84925
150, 150, 30, 16 .85025
150, 150, 30, 300 .8595, 0.836
150, 150, 45, 400 .851
150, 150, 8 .862, .85275, .8445, .8495, .847, .847, .8495, .834, .8465, 0.84775
150, 150, 10 .86025, .84425, .856, .85825, .852
150, 150, 11 .85175, .841, .85
150, 150, 12 .85175, .85175, .848
150, 150, 14 .86425, .8575
150, 150, 15 .851
150, 200, 10 .84925, .84875
160, 160, 10 .85975, 0.8485, .85325, 0.8445
160, 160, 12 .84
180, 180, 10 .85125, 0.846
180, 180, 12 .852, 0.84975
180, 180, 14 .848
200, 200, 8 .84475, .84725, .845
200, 200, 10 .85425, .84925, .8595, .8505, .847
200, 200, 12 .84875, .84675, .85375
more nodes = less epochs
leaky relu
12 epochs
100, 100 .8475
150, 150 .856, .83225, .84825, .841, .848
200, 200 .85825, .844, .85775, .859, .84925
300, 300 .84925
alpha = .03 .851 -reverted
sigmoid -> leaky .8375 -rev
add sigmoid layer .8535 -rev
delete dropout .8515, .848
alpha = .3 .85625
add batchnorm .852
epochs = 8 .85175, .85225, .852, .856, .85225, .85525, .85325
epochs = 10 .847
epochs = 9 .84975
epochs = 11 .849
'''