def get_theta_of_problem(matrix, nr_features, nr_epochs, nr_iterations):
nr_problems, nr_users = matrix.shape
theta_users = np.random.rand(nr_users, nr_features)
theta_problems = np.zeros((nr_problems, nr_features))
matrix_t = np.transpose(matrix)
i = 0
while i <= nr_epochs:
print('Progress: {}%'.format(100 * i / nr_epochs))
i += 1
for problem in range(nr_problems):
x = np.array(theta_users)
y = np.array(matrix[problem])
theta_problems[problem] = BaseRecommender.gradient_descent(
x, y, nr_iterations)[1:]
for user in range(nr_users):
x = np.array(theta_problems)
y = np.array(matrix_t[user])
theta_users[user] = BaseRecommender.gradient_descent(
x, y, nr_iterations)[1:]
return theta_problems.mean(axis=0)
def __init__(self, users=None, items=None, config=None):
BaseRecommender.__init__(self, users, items, config)
self.P = None
self.Q = None
self.b_i = None
if config and self.users and self.items:
nb_users = len(self.users)
nb_movies = len(self.items)
nb_latent_f = config['nb_latent_f']
scale = config['init_params_scale'] if 'init_params_scale' in config else 0.001
params = self._init_params(nb_users, nb_movies, nb_latent_f, scale=scale)
self._set_params(params)
def __init__(self, users=None, items=None, config=None):
BaseRecommender.__init__(self, users, items, config)
self.sparse_matrix = None
self.similarity_matrix = None
self.intercepts = None
if config:
self.ignore_negative_weights = config['ignore_negative_weights']
self.model = SGDRegressor(penalty='elasticnet', n_iter=config['nb_epochs'],
fit_intercept=config['fit_intercept'], alpha=config['alpha'],
l1_ratio=config['l1_ratio'])
self.intercepts = {}
def __init__(self, users=None, items=None, config=None, movie_to_imdb=None, user_pref_model=None, d2v_model=None):
BaseRecommender.__init__(self, users, items, config)
self.user_factors = None
self.item_factors = None
self.item_bias = None
self.global_bias = None
self.user_pref_model = user_pref_model
self.d2v_model = d2v_model
self.movie_to_imdb = movie_to_imdb
if config and self.users and self.items:
nb_users = len(self.users)
nb_movies = len(self.items)
nb_latent_f = config['nb_latent_f']
params = self._init_params(nb_users, nb_movies, nb_latent_f)
self._set_params(params)
def load_recommender(filepath):
"""
Load a recommender model from file after it has been saved by
save_recommender().
Parameters
----------
filepath : str
The filepath to read from.
"""
return BaseRecommender.load(filepath)
def load_recommender(filepath):
"""
Load a recommender model from file after it has been saved by
save_recommender().
Parameters
----------
filepath : str
The filepath to read from.
"""
return BaseRecommender.load(filepath)
def read_recommender_description(filepath):
"""
Read a recommender model description from file after it has
been saved by save_recommender(), without loading all the
associated data into memory.
Parameters
----------
filepath : str
The filepath to read from.
"""
return BaseRecommender.read_recommender_description(filepath)
def __init__(self, users=None, items=None, config=None):
BaseRecommender.__init__(self, users, items, config)
self.user_factors = None
self.item_factors = None
self.user_interest_factors = None
self.user_interest_bias = None
self.item_bias = None
self.global_bias = None
self.si_user_model = None
self.si_item_model = None
if config and self.users and self.items:
nb_users = len(self.users)
nb_movies = len(self.items)
nb_latent_f = config['nb_latent_f']
nb_user_pref = config['nb_user_pref']
scale = config['init_params_scale'] if 'init_params_scale' in config else 0.001
params = self._init_params(nb_users, nb_movies, nb_latent_f, nb_user_pref, scale=scale)
self._set_params(params)
def read_recommender_description(filepath):
"""
Read a recommender model description from file after it has
been saved by save_recommender(), without loading all the
associated data into memory.
Parameters
----------
filepath : str
The filepath to read from.
"""
return BaseRecommender.read_recommender_description(filepath)
def __init__(self, users=None, items=None, config=None, movie_to_imdb=None,d2v_model=None):
BaseRecommender.__init__(self, users, items, config)
self.user_factors = None
self.item_factors = None
self.item_bias = None
self.global_bias = None
self.nn_w1 = None
self.nn_w2 = None
self.d2v_model = d2v_model
self.movie_to_imdb = movie_to_imdb
if config and self.users and self.items:
nb_users = len(self.users)
nb_movies = len(self.items)
self.nb_latent_f = config['nb_latent_f']
nb_d2v_features = config['nb_d2v_features']
nb_hidden_neurons = config['nb_hidden_neurons']
params = self._init_params(nb_users, nb_movies, self.nb_latent_f, nb_d2v_features, nb_hidden_neurons)
self._set_params(params)
