def test_predict_item_repr_biased_fails(self):
model = TensorRec(n_components=10)
model.fit(self.interactions,
self.user_features,
self.item_features,
epochs=10)
with self.assertRaises(NotImplementedError):
model.predict_item_representation(self.item_features)
def test_predict_item_repr(self):
model = TensorRec(n_components=10, biased=False)
model.fit(self.interactions,
self.user_features,
self.item_features,
epochs=10)
item_repr = model.predict_item_representation(self.item_features)
self.assertEqual(item_repr.shape, (self.item_features.shape[0], 10))
def test_predict_fail_unfit(self):
model = TensorRec()
with self.assertRaises(ModelNotFitException):
model.predict(self.user_features, self.item_features)
with self.assertRaises(ModelNotFitException):
model.predict_rank(self.user_features, self.item_features)
with self.assertRaises(ModelNotFitException):
model.predict_user_representation(self.user_features)
with self.assertRaises(ModelNotFitException):
model.predict_item_representation(self.item_features)
with self.assertRaises(ModelNotFitException):
model.predict_user_attention_representation(self.user_features)
with self.assertRaises(ModelNotFitException):
model.predict_similar_items(self.item_features,
item_ids=[1],
n_similar=5)
with self.assertRaises(ModelNotFitException):
model.predict_item_bias(self.item_features)
with self.assertRaises(ModelNotFitException):
model.predict_user_bias(self.user_features)
normalize_users=True,
normalize_items=True,
n_tastes=3)
# Make some random selections of movies and users we want to plot
movies_to_plot = np.random.choice(a=item_features.shape[0], size=200, replace=False)
user_to_plot = np.random.choice(a=user_features.shape[0], size=400, replace=False)
# Iterate through 1000 epochs, outputting a JPG plot each epoch
for epoch in range(epochs):
model.fit_partial(interactions=train_interactions, user_features=user_features, item_features=item_features,
**fit_kwargs)
# The position of a movie or user is that movie's/user's 2-dimensional representation. The size of the movie dot is
# related to its item bias.
movie_positions = model.predict_item_representation(item_features)
user_positions = model.predict_user_representation(user_features)
movie_sizes = model.predict_item_bias(item_features) * 10 + 1.0
# Handle multiple tastes, if applicable. If there are more than 1 taste per user, only the first of each user's
# tastes will be plotted.
if model.n_tastes > 1:
user_positions = user_positions[0]
_, ax = plt.subplots()
ax.grid(b=True, which='both')
ax.axhline(y=0, color='k')
ax.axvline(x=0, color='k')
ax.scatter(*zip(*user_positions[user_to_plot]), color='r', s=1)
ax.scatter(*zip(*movie_positions[movies_to_plot]), s=movie_sizes)
ax.set_aspect('equal')
user_to_plot = np.random.choice(a=user_features.shape[0], size=100, replace=False)
# Coerce data to datasets for faster fitting
train_interactions_ds = create_tensorrec_dataset_from_sparse_matrix(train_interactions)
user_features_ds = create_tensorrec_dataset_from_sparse_matrix(user_features)
item_features_ds = create_tensorrec_dataset_from_sparse_matrix(item_features)
# Iterate through 1000 epochs, outputting a JPG plot each epoch
for epoch in range(epochs):
model.fit_partial(interactions=train_interactions_ds,
user_features=user_features_ds,
item_features=item_features_ds,
**fit_kwargs)
# The position of a movie or user is that movie's/user's 2-dimensional representation.
movie_positions = model.predict_item_representation(item_features_ds)
user_positions = model.predict_user_representation(user_features_ds)
# Handle multiple tastes, if applicable. If there are more than 1 taste per user, only the first of each user's
# tastes will be plotted.
if model.n_tastes > 1:
user_positions = user_positions[0]
_, ax = plt.subplots()
ax.grid(b=True, which='both')
ax.axhline(y=0, color='k')
ax.axvline(x=0, color='k')
ax.scatter(*zip(*user_positions[user_to_plot]), color='r', s=1)
ax.scatter(*zip(*movie_positions[movies_to_plot]), s=2)
ax.set_aspect('equal')