def shuffle_in_unison(a, b):
assert len(a) == len(b)
shuffled_a = np.empty(a.shape, dtype=a.dtype)
shuffled_b = np.empty(b.shape, dtype=b.dtype)
permutation = np.random.permutation(len(a))
for old_index, new_index in enumerate(permutation):
shuffled_a[new_index] = a[old_index]
shuffled_b[new_index] = b[old_index]
return shuffled_a, shuffled_b
if __name__ == "__main__":
# print(1)
X_train, X_test, y_train, y_test = prepare_data(m1matrix_2)
X_train, y_train = shuffle_in_unison(X_train, y_train)
array2file(X_train, y_train, redundancy_output_dir + '\\svm\\m1_train_2.data')
array2file(X_test, y_test, redundancy_output_dir + '\\svm\\m1_test_2.data')
X_train, X_test, y_train, y_test = prepare_data(m2matrix_2)
X_train, y_train = shuffle_in_unison(X_train, y_train)
array2file(X_train, y_train, redundancy_output_dir + '\\svm\\m2_train_2.data')
array2file(X_test, y_test, redundancy_output_dir + '\\svm\\m2_test_2.data')
X_train, X_test, y_train, y_test = prepare_data(m4matrix_2)
X_train, y_train = shuffle_in_unison(X_train, y_train)
array2file(X_train, y_train, redundancy_output_dir + '\\svm\\m4_train_2.data')
array2file(X_test, y_test, redundancy_output_dir + '\\svm\\m4_test_2.data')
X_train, X_test, y_train, y_test = prepare_data(m8matrix_2)
X_train, y_train = shuffle_in_unison(X_train, y_train)
array2file(X_train, y_train, redundancy_output_dir + '\\svm\\m8_train_2.data')
abundant_id_list = np.load(os.path.join(tensorflow_data_3_dir, 'top250_movie_pos.npy'))
X_test = result_matrix[abundant_id_list,]
y_test = movie_rates_y[abundant_id_list,]
X_train = np.delete(result_matrix, abundant_id_list, 0)
y_train = np.delete(movie_rates_y, abundant_id_list, 0)
print(X_train.shape, y_train.shape)
array2file(X_train, y_train, SVD3_output_dir + '\\NeuMF\\NeuMF_train_{}.data'.format(n))
array2file(X_test, y_test, SVD3_output_dir + '\\NeuMF\\NeuMF_test_{}.data'.format(n))
print('NeuMF extension finish')
'''
# MLP
matrix = np.load(os.path.join(workdir, 'SVD_3', 'baselines', 'MLP_result_mt_ori.npy'))
movie_rates_y = np.load(os.path.join(tensorflow_data_3_dir, "movie_rates.npy"))
for n in range(5, 25, 5):
result_matrix, extension_matrix, sort_index_matrix = tag_extension_2(matrix, n)
abundant_id_list = np.load(os.path.join(tensorflow_data_3_dir, 'top250_movie_pos.npy'))
X_test = result_matrix[abundant_id_list,]
y_test = movie_rates_y[abundant_id_list,]
X_train = np.delete(result_matrix, abundant_id_list, 0)
y_train = np.delete(movie_rates_y, abundant_id_list, 0)
print(X_train.shape, y_train.shape)
array2file(X_train, y_train, SVD3_output_dir + '\\MLP\\MLP_train_{}.data'.format(n))
array2file(X_test, y_test, SVD3_output_dir + '\\MLP\\MLP_test_{}.data'.format(n))
print('MLP extension finish')
all_movie_pos_file_Path = os.path.join(workdir,
"all_movie_matrix_dict.json")
all_movie_pos_file = open(all_movie_pos_file_Path,
encoding='utf-8',
mode='r')
all_movie_pos_dict = json.load(all_movie_pos_file)
most_popular_file_Path = os.path.join(redundancy_output_dir,
"most_popular_movie.json")
most_popular_file = open(most_popular_file_Path,
encoding='utf-8',
mode='r')
most_popular_movie_dict = json.load(most_popular_file)
most_popular_movie_list = most_popular_movie_dict.keys()
X_train, X_test, y_train, y_test = prepare_data(matrix, movie_rates_y,
most_popular_movie_list,
all_movie_pos_dict)
X_train, y_train = shuffle_in_unison(X_train, y_train)
array2file(X_train, y_train,
SVD_output_dir + '\\svm\\k{}_train.data'.format(k))
array2file(X_test, y_test,
SVD_output_dir + '\\svm\\k{}_test.data'.format(k))
tagFile = os.path.join(redundancy_output_dir, 'tag.txt')
tag_list = loadTagList(tagFile)
SVD_movie_result = os.path.join(SVD_output_dir,
"SVD_movie_result_{}.json".format(k))
matrix2json(extension_matrix, all_movie_pos_dict, tag_list,
SVD_movie_result)
y_test = np.delete(y, id_list, 0)
print(X_train.shape, y_train.shape)
return X_train, y_train, X_test, y_test
def shuffle_in_unison(a, b):
assert len(a) == len(b)
shuffled_a = np.empty(a.shape, dtype=a.dtype)
shuffled_b = np.empty(b.shape, dtype=b.dtype)
permutation = np.random.permutation(len(a))
for old_index, new_index in enumerate(permutation):
shuffled_a[new_index] = a[old_index]
shuffled_b[new_index] = b[old_index]
return shuffled_a, shuffled_b
if __name__ == "__main__":
# print(1)
# X_train, X_test, y_train, y_test = prepare_data(m1matrix_2)
# X_train, y_train = shuffle_in_unison(X_train, y_train)
# array2file(X_train, y_train, redundancy_output_dir + '\\svm\\m1_train_2.data')
# array2file(X_test, y_test, redundancy_output_dir + '\\svm\\m1_test_2.data')
print("prepare r1matrix data")
abundant_movie_matrix = os.path.join(redundancy2_output_dir, 'r1matrix_2.npy')
abundant_movie_rates = os.path.join(tensorflow_data_dir, 'abundant_movie_rates.npy')
X_train, y_train, X_test, y_test = get_60_split(abundant_movie_matrix, abundant_movie_rates)
array2file(X_train, y_train, redundancy2_output_dir + '\\svm\\m1_train_addi.data')
array2file(X_test, y_test, redundancy2_output_dir + '\\svm\\m1_test_addi.data')