/
Experiment2.py
200 lines (168 loc) · 10.5 KB
/
Experiment2.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
__author__ = 'frankhe'
import subprocess
from Generate import compute_error
from Generate import compute_classification
from libFM_tool import DataProcess
import random
dataProcess = DataProcess()
def baseline1_random(active_learning_ratio=0.1):
print '\n==================================='
print 'baseline1 with random choosing active learning data started'
print '==================================='
""" Choose active learning data randomly """
number_of_active_data = (int(dataProcess.movieDataBase.TOTAL_USERS*active_learning_ratio)-1)*240
number_of_index = len(dataProcess.test_addAllNegative_data)
selected_data_positions = set()
while len(selected_data_positions) < number_of_active_data:
selected_data_positions.add(random.randint(0, number_of_index-1))
alternative_user_movie_list = []
for index in selected_data_positions:
userId = dataProcess.test_addAllNegative_data[index][0]
movieId = dataProcess.test_addAllNegative_data[index][1]
alternative_user_movie_list.append([userId, movieId])
# movieDataBase.make_alternative_user_movie_matrix(alternative_user_movie_list)
""" add active learning result into train_original_data """
print '\n==================================='
print 'active learning regression started'
print '==================================='
dataProcess.movieDataBase.make_user_movie_rating_matrix(dataProcess.test_original_data)
active_learning_train_data = []
# this scheme is adding every thing in active learning
count = 0
for values in alternative_user_movie_list:
userId = values[0]
movieId = values[1]
ratings = dataProcess.movieDataBase.user_movie_rating_matrix.get(userId)
if ratings is not None:
rating = ratings.get(movieId)
if rating is not None:
active_learning_train_data.append([userId, movieId, rating])
count += 1
train_add_active_learning_data = dataProcess.train_original_data + active_learning_train_data
dataProcess.movieDataBase.store_data_to_file(train_add_active_learning_data, fileName='train_add_active_learning_data')
dataProcess.movieDataBase.generate_libfm_data(train_add_active_learning_data)
dataProcess.movieDataBase.store_data_to_file(dataProcess.movieDataBase.libfm_data, fileName='train_step3.libfm')
# subprocess.call("./Generate/libFM -task r -train Generate/train_step3.libfm -test Generate/test_step3.libfm "
# "-method mcmc -out Generate/prediction", shell=True)
# compute_error.computer_error()
subprocess.call("./Generate/libFM -task r -train Generate/train_step3.libfm -test Generate/test_step3.libfm "
"-method sgd -dim '1,1, 200' -learn_rate 0.001 -iter 30 -out Generate/prediction", shell=True)
print 'number of alternative user-movie requests=', len(alternative_user_movie_list)
print 'number of gained active learning user-movie data=', count
compute_error.computer_error()
def baseline2_random_after_classification(active_learning_ratio=0.1):
print '\n==================================='
print 'baseline2 random choosing active learning data after classification started'
print '==================================='
"""the test data is still in movieDataBase.core. Next is the step 1 """
print '\n==================================='
print 'step 1 binary classification started'
print '==================================='
subprocess.call("./Generate/libFM -task c -train Generate/train_step1.libfm -test Generate/test_step1.libfm "
"-method mcmc -iter 1 -out Generate/prediction", shell=True)
# subprocess.call("./Generate/libFM -task c -train Generate/train_step1.libfm -test Generate/test_step1.libfm "
# "-method sgd -learn_rate 0.01 -out Generate/prediction", shell=True)
selected_data_positions_tmp = compute_classification.compute_classification(len(dataProcess.test_original_data))
number_of_active_data = (int(dataProcess.movieDataBase.TOTAL_USERS*active_learning_ratio)-1)*240
number_of_index = len(dataProcess.test_addAllNegative_data)
selected_data_positions = set()
while len(selected_data_positions) < number_of_active_data:
""" next line is wrong but has good experiment behavior i want to know why."""
# selected_data_positions.add(random.randint(0, len(selected_data_positions_tmp)-1))
selected_data_positions.add(random.choice(selected_data_positions_tmp))
alternative_user_movie_list = []
for index in selected_data_positions:
userId = dataProcess.test_addAllNegative_data[index][0]
movieId = dataProcess.test_addAllNegative_data[index][1]
alternative_user_movie_list.append([userId, movieId])
# movieDataBase.make_alternative_user_movie_matrix(alternative_user_movie_list)
""" add active learning result into train_original_data """
print '\n==================================='
print 'step active learning regression started'
print '==================================='
dataProcess.movieDataBase.make_user_movie_rating_matrix(dataProcess.test_original_data)
active_learning_train_data = []
# this scheme is adding every thing in active learning
count = 0
for values in alternative_user_movie_list:
userId = values[0]
movieId = values[1]
ratings = dataProcess.movieDataBase.user_movie_rating_matrix.get(userId)
if ratings is not None:
rating = ratings.get(movieId)
if rating is not None:
active_learning_train_data.append([userId, movieId, rating])
count += 1
train_add_active_learning_data = dataProcess.train_original_data + active_learning_train_data
dataProcess.movieDataBase.store_data_to_file(train_add_active_learning_data, fileName='train_add_active_learning_data')
dataProcess.movieDataBase.generate_libfm_data(train_add_active_learning_data)
dataProcess.movieDataBase.store_data_to_file(dataProcess.movieDataBase.libfm_data, fileName='train_step3.libfm')
# subprocess.call("./Generate/libFM -task r -train Generate/train_step3.libfm -test Generate/test_step3.libfm "
# "-method mcmc -out Generate/prediction", shell=True)
# compute_error.computer_error()
subprocess.call("./Generate/libFM -task r -train Generate/train_step3.libfm -test Generate/test_step3.libfm "
"-method sgd -dim '1,1, 200' -learn_rate 0.001 -iter 120 -out Generate/prediction", shell=True)
print 'number of alternative user-movie requests=', len(alternative_user_movie_list)
print 'number of gained active learning user-movie data=', count
compute_error.computer_error()
def experiment2_user_qualification(active_learning_ratio=0.1):
print '\n==================================='
print 'experiment2 choosing qualified active learning data after classification started'
print '==================================='
"""the test data is still in movieDataBase.core. Next is the step 1 """
print '\n==================================='
print 'step 1 binary classification started'
print '==================================='
subprocess.call("./Generate/libFM -task c -train Generate/train_step1.libfm -test Generate/test_step1.libfm "
"-method mcmc -iter 1 -out Generate/prediction", shell=True)
# subprocess.call("./Generate/libFM -task c -train Generate/train_step1.libfm -test Generate/test_step1.libfm "
# "-method sgd -learn_rate 0.01 -out Generate/prediction", shell=True)
selected_data_positions = compute_classification.compute_classification(len(dataProcess.test_original_data))
print '\n==================================='
print 'choosing users with low rating MSE'
print '==================================='
number_of_active_data = (int(dataProcess.movieDataBase.TOTAL_USERS*active_learning_ratio)-1)*240
number_of_index = len(dataProcess.test_addAllNegative_data)
alternative_user_movie_list = []
for index in selected_data_positions:
userId = dataProcess.test_addAllNegative_data[index][0]
movieId = dataProcess.test_addAllNegative_data[index][1]
alternative_user_movie_list.append([userId, movieId])
alternative_user_movie_list = sorted(alternative_user_movie_list, key=lambda x: dataProcess.movieDataBase.user_MSE[x[0]])
alternative_user_movie_list = alternative_user_movie_list[:number_of_active_data]
# for x, y in alternative_user_movie_list:
# print x, y
# movieDataBase.make_alternative_user_movie_matrix(alternative_user_movie_list)
""" add active learning result into train_original_data """
print '\n==================================='
print 'step active learning regression started'
print '==================================='
dataProcess.movieDataBase.make_user_movie_rating_matrix(dataProcess.test_original_data)
active_learning_train_data = []
# this scheme is adding every thing in active learning
count = 0
for values in alternative_user_movie_list:
userId = values[0]
movieId = values[1]
ratings = dataProcess.movieDataBase.user_movie_rating_matrix.get(userId)
if ratings is not None:
rating = ratings.get(movieId)
if rating is not None:
active_learning_train_data.append([userId, movieId, rating])
count += 1
train_add_active_learning_data = dataProcess.train_original_data + active_learning_train_data
dataProcess.movieDataBase.store_data_to_file(train_add_active_learning_data, fileName='train_add_active_learning_data')
dataProcess.movieDataBase.generate_libfm_data(train_add_active_learning_data)
dataProcess.movieDataBase.store_data_to_file(dataProcess.movieDataBase.libfm_data, fileName='train_step3.libfm')
# subprocess.call("./Generate/libFM -task r -train Generate/train_step3.libfm -test Generate/test_step3.libfm "
# "-method mcmc -out Generate/prediction", shell=True)
# compute_error.computer_error()
subprocess.call("./Generate/libFM -task r -train Generate/train_step3.libfm -test Generate/test_step3.libfm "
"-method sgd -dim '1,1, 200' -learn_rate 0.001 -iter 50 -out Generate/prediction", shell=True)
print 'number of alternative user-movie requests=', len(alternative_user_movie_list)
print 'number of gained active learning user-movie data=', count
compute_error.computer_error()
if __name__ == '__main__':
# baseline1_random()
# baseline2_random_after_classification()
experiment2_user_qualification()