-
Notifications
You must be signed in to change notification settings - Fork 0
/
final_model_66.2.py
375 lines (229 loc) · 11.8 KB
/
final_model_66.2.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
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
# coding: utf-8
# In[1]:
import numpy as np
import math
import pandas as pd
from tqdm import tqdm
import keras
from keras.utils import np_utils
from keras.models import Model
from keras.layers import Input, Dense, Dropout, Embedding, Flatten
from keras.callbacks import ModelCheckpoint
from keras.initializers import RandomUniform
import cPickle as pickle
import progressbar
import os
# In[2]:
def load_dataset():
return pd.read_csv('../Data/train.csv').fillna('')
# In[3]:
def load_songs_members_dataset():
songs_dataset = pd.read_csv('../../new_data/Data/songs.csv')
members_dataset = pd.read_csv('../Data/members.csv')
return songs_dataset, members_dataset
# In[4]:
# In[5]:
# Some of the rows corresponding to a column have multiple values separated by '|'
# character. We need to split and separate these multiple values
def create_mapper(values):
mapper = dict()
for v in values:
mapper[v] = len(v)
return mapper
# In[6]:
def load_song_mapper_picke():
song_mapper = pickle.load(open('../New_Data/model_song_embeddings/song_mapper_py2.pkl', 'rb'))
return song_mapper
def load_user_mapper_pickle():
user_mapper = pickle.load(open('../New_Data/model_user_embeddings/msno_mapper_py2.pkl', 'rb'))
return user_mapper
# In[7]:
train_dataset = load_dataset()
songs_dataset, members_dataset = load_songs_members_dataset()
# In[87]:
train_dataset = train_dataset.merge(members_dataset, on='msno', how='left')
# In[88]:
train_dataset = train_dataset.merge(songs_dataset, on='song_id', how='left')
# In[89]:
def genre_id_count(x):
if x == 'no_genre_id':
return 0
else:
return x.count('|') + 1
train_dataset['genre_ids'].fillna('no_genre_id',inplace=True)
train_dataset['genre_ids_count'] = train_dataset['genre_ids'].apply(genre_id_count).astype(np.int32)
train_dataset['genre_ids_count'] = train_dataset['genre_ids_count'] / max(train_dataset['genre_ids_count'])
# In[90]:
def artist_count(x):
if x == 'no_artist':
return 0
else:
return x.count('and') + x.count(',') + x.count('feat') + x.count('&')
train_dataset['artist_name'].fillna('no_artist',inplace=True)
train_dataset['artist_count'] = train_dataset['artist_name'].map(str).apply(artist_count).astype(np.int32)
train_dataset['artist_count'] = train_dataset['artist_count'] / max(train_dataset['artist_count'])
# In[91]:
def lyricist_count(x):
if x == 'no_lyricist':
return 0
else:
return sum(map(x.count, ['|', '/', '\\', ';'])) + 1
return sum(map(x.count, ['|', '/', '\\', ';']))
train_dataset['lyricist'].fillna('no_lyricist',inplace=True)
train_dataset['lyricists_count'] = train_dataset['lyricist'].map(str).apply(lyricist_count).astype(np.int32)
train_dataset['lyricists_count'] = train_dataset['lyricists_count'] / max(train_dataset['lyricists_count'])
# In[92]:
def composer_count(x):
if x == 'no_composer':
return 0
else:
return sum(map(x.count, ['|', '/', '\\', ';'])) + 1
train_dataset['composer'].fillna('no_composer',inplace=True)
train_dataset['composer_count'] = train_dataset['composer'].map(str).apply(composer_count).astype(np.int8)
train_dataset['composer_count'] = train_dataset['composer_count'] / max(train_dataset['composer_count'])
# In[93]:
# number of times a song has been played before
_dict_count_song_played_train = {k: v for k, v in train_dataset['song_id'].value_counts().iteritems()}
def count_song_played(x):
try:
return _dict_count_song_played_train[x]
except KeyError:
return 0
train_dataset['count_song_played'] = train_dataset['song_id'].map(str).apply(count_song_played).astype(np.int64)
train_dataset['count_song_played'] = train_dataset['count_song_played'] / max(train_dataset['count_song_played'])
# In[94]:
# number of times an artist has been played
_dict_count_artist_played_train = {k: v for k, v in train_dataset['artist_name'].value_counts().iteritems()}
def count_artist_played(x):
try:
return _dict_count_artist_played_train[x]
except KeyError:
return 0
train_dataset['count_artist_played'] = train_dataset['artist_name'].map(str).apply(count_artist_played).astype(np.int64)
train_dataset['count_artist_played'] = train_dataset['count_artist_played'] / max(train_dataset['count_artist_played'])
source_tab_mapper = create_mapper(train_dataset.source_system_tab.unique())
s_scr_name_mapper = create_mapper(train_dataset.source_screen_name.unique())
s_type_mapper = create_mapper(train_dataset.source_type.unique())
msno_mapper = load_user_mapper_pickle()
song_id_mapper = load_song_mapper_picke()
# In[8]:
def get_model():
input_song_ids_layer = Input(shape=(1,))
input_msno_layer = Input(shape=(1, ))
input_source_system_tab = Input(shape=(1, ))
input_source_screen_name = Input(shape=(1, ))
input_source_type = Input(shape=(1, ))
input_genre_ids_count = Input(shape=(1, ))
input_artist_count = Input(shape=(1, ))
input_lyricist_count = Input(shape=(1, ))
input_composer_count = Input(shape=(1, ))
input_song_count_played = Input(shape=(1, ))
input_artist_count_played = Input(shape=(1, ))
user_embeddings = Embedding(input_dim = len(msno_mapper) + 1,
output_dim = 64,
embeddings_initializer=RandomUniform(minval=-0.5, maxval=0.5),
embeddings_regularizer=keras.regularizers.l2(1e-4),
input_length=1,
trainable=True)(input_msno_layer)
song_embeddings = Embedding(input_dim=len(song_id_mapper) + 1, output_dim=64,
embeddings_initializer=RandomUniform(minval=-0.5, maxval=0.5),
embeddings_regularizer=keras.regularizers.l2(1e-4), input_length=1, trainable=True)(input_song_ids_layer)
flatten_user = Flatten()(user_embeddings)
flatten_songs = Flatten()(song_embeddings)
sys_tab_embeddings = Embedding(output_dim=10, input_dim= max(source_tab_mapper.values()) + 1,input_length=1, trainable=True, embeddings_regularizer=keras.regularizers.l2(1e-3), embeddings_initializer='glorot_uniform') (input_source_system_tab)
screen_name_embeddings = Embedding(output_dim=10, input_dim= max(s_scr_name_mapper.values()) + 1,input_length=1, trainable=True, embeddings_regularizer=keras.regularizers.l2(1e-3),embeddings_initializer='glorot_uniform' ) (input_source_screen_name)
source_type_embeddings = Embedding(output_dim=10, input_dim= max(s_type_mapper.values()) + 1,input_length=1, trainable=True, embeddings_regularizer=keras.regularizers.l2(1e-3), embeddings_initializer='glorot_uniform') (input_source_type)
flatten_tabs = Flatten()(sys_tab_embeddings)
flatten_screen = Flatten()(screen_name_embeddings)
flatten_type = Flatten()(source_type_embeddings)
combined_meta = keras.layers.concatenate([flatten_tabs, flatten_type, flatten_screen])
combined_counts = keras.layers.concatenate([input_genre_ids_count, input_artist_count, input_lyricist_count, input_composer_count, input_song_count_played, input_artist_count_played])
dense_counts = Dense(40, kernel_initializer='glorot_uniform', bias_initializer='zeros', activation='relu')(combined_counts)
dense_meta = Dense(40, kernel_initializer='glorot_uniform', bias_initializer='zeros', activation='relu')(combined_meta)
dot_user_songs = keras.layers.dot([flatten_user, flatten_songs], axes = 1)
combined_input = keras.layers.concatenate([flatten_user, flatten_songs, dot_user_songs, dense_meta, dense_counts])
intermediate_0 = Dense(128, kernel_initializer='glorot_uniform', bias_initializer='zeros', activation='relu')(combined_input)
dropout_0 = Dropout(0.5)(intermediate_0)
intermediate_1 = Dense(64, kernel_initializer='glorot_uniform', bias_initializer='zeros', activation='relu')(dropout_0)
dropout_1 = Dropout(0.5)(intermediate_1)
intermediate_2 = Dense(16, kernel_initializer='glorot_uniform', bias_initializer='zeros', activation='relu')(dropout_1)
output_0 = Dense(1, kernel_initializer='glorot_uniform', activation='sigmoid')(intermediate_2)
dnn_model = keras.models.Model(inputs = [input_msno_layer, input_song_ids_layer, input_source_system_tab, input_source_screen_name, input_source_type, input_genre_ids_count, input_artist_count, input_lyricist_count, input_composer_count, input_song_count_played, input_artist_count_played],
outputs = [output_0])
dnn_model.compile(optimizer=keras.optimizers.Adam(lr=1e-4), loss='binary_crossentropy', metrics=['accuracy'])
return dnn_model
# In[9]:
dnn_model = get_model()
# In[10]:
print dnn_model.summary()
# In[11]:
from IPython.display import SVG
from keras.utils.vis_utils import model_to_dot
SVG(model_to_dot(dnn_model).create(prog='dot', format='svg'))
# In[12]:
# In[13]:
def gen_train_data(data):
num_rows = data.shape[0]
X_msno = np.empty(num_rows)
X_song_id = np.empty(num_rows)
X_source_system_tab = np.empty(num_rows)
X_source_screen_name = np.empty(num_rows)
X_source_type = np.empty(num_rows)
Y = np.empty(num_rows)
batch = num_rows
input_genre_ids_count = np.empty(batch)
input_artist_count = np.empty(batch)
input_lyricist_count = np.empty(batch)
input_composer_count = np.empty(batch)
input_song_count_played = np.empty(batch)
input_artist_count_played = np.empty(batch)
count = 0
bar = progressbar.ProgressBar()
for _, row in bar(data.iterrows()):
X_msno[count,] = msno_mapper[row['msno']]
X_song_id[count, ] = song_id_mapper[row['song_id']]
X_source_system_tab[count, ] = source_tab_mapper[row['source_system_tab']]
X_source_screen_name[count, ] = s_scr_name_mapper[row['source_screen_name']]
X_source_type[count, ] = s_type_mapper[row['source_type']]
Y[count] = row['target']
input_genre_ids_count[count, ] = row['genre_ids_count']
input_artist_count[count, ] = row['artist_count']
input_lyricist_count[count, ] = row['lyricists_count']
input_composer_count[count, ] = row['composer_count']
input_song_count_played[count, ] = row['count_song_played']
input_artist_count_played[count, ] = row['count_artist_played']
count += 1
return ([X_msno, X_song_id, X_source_system_tab, X_source_screen_name, X_source_type, input_genre_ids_count, input_artist_count, input_lyricist_count, input_composer_count, input_song_count_played, input_artist_count_played], [Y])
# In[14]:
dst = train_dataset[train_dataset.song_id != 'lLsg/q8lurYYpIc3X526qxpWD6SY8Y4grHSCfXlO3mM=']
dst = dst[train_dataset.song_id != '5FXnI1sbD+lFFFGULwZhUAbZMa1P1eFMiigTjMZgW0I=']
train_dataset = dst
# In[15]:
mask_val = 0.8
len2 = len(train_dataset) * mask_val
len2 = int(len2)
temp = train_dataset
train_data = temp[0:len2]
val_data = temp[len2:len(temp)]
print 'Train Data --> ', len(train_data)
print 'Val Data --> ', len(val_data)
# tr = gen_train_data(train_data)
# In[16]:
trX, trY = gen_train_data(train_data)
valX, valY = gen_train_data(val_data)
# In[ ]:
# filepath="weights-improvement-{epoch:02d}-{val_acc:.2f}.hdf5"
# checkpoint = ModelCheckpoint(filepath, monitor='val_acc', verbose=0, save_best_only=True, mode='max')
# callbacks_list = [checkpoint]
# In[ ]:
batch_size = 32768
# generator = input_generator(train_data, batch_size)
# val_generator = input_generator(val_data, batch_size)
# print 'total iterations -- ' , len(train_data)/batch_size
dnn_model.fit(x=trX, y=trY, batch_size = batch_size, epochs = 100, verbose=2, validation_data=(valX, valY),
shuffle=True)
# dnn_model.fit_generator(generator=generator, steps_per_epoch = (len(train_data)/batch_size), validation_data = val_generator, validation_steps = (len(val_data)/batch_size), epochs=2)
# In[ ]:
model_file = 'lysto_model_no_pre_train_embeds_md_counts_v1_epochs_100.h5'
dnn_model.save(model_file)
print model_file, 'Model saved'