def test_get_item_feature_with_new_items(self):
n_users = 101
n_items = 233
n_data = 3007
am1 = _make_sparse_matrix(n_users, n_items, n_data)
am2 = 2 * _make_sparse_matrix(n_users, n_items, n_data)
adjacency_matrix = am1 + am2
user_ids = adjacency_matrix.tocoo().row
item_ids = adjacency_matrix.tocoo().col
ratings = adjacency_matrix.tocoo().data
item_features = [{i: np.array([i]) for i in range(n_items)}]
dataset = GcmcDataset(user_ids,
item_ids,
ratings,
item_features=item_features)
graph_dataset = GcmcGraphDataset(dataset, test_size=0.1)
encoder_hidden_size = 100
encoder_size = 100
scope_name = 'GraphConvolutionalMatrixCompletionGraph'
model = GraphConvolutionalMatrixCompletion(
graph_dataset=graph_dataset,
encoder_hidden_size=encoder_hidden_size,
encoder_size=encoder_size,
scope_name=scope_name,
batch_size=1024,
epoch_size=10,
learning_rate=0.01,
dropout_rate=0.7,
normalization_type='symmetric')
model.fit()
user_ids = [90, 62, 3, 3]
item_ids = [11, 236, 240, 243]
additional_item_features = {
item_id: np.array([999])
for item_id in item_ids
}
additional_dataset = GcmcDataset(
np.array(user_ids),
np.array(item_ids),
np.array([1, 2, 1, 1]),
item_features=[additional_item_features])
target_item_ids = item_ids + [12, 13, 17, 55
] # item_ids to get embeddings
item_feature = model.get_item_feature_with_new_items(
item_ids=target_item_ids, additional_dataset=additional_dataset)
self.assertEqual(len(item_feature), 2)
self.assertEqual(list(item_feature[0]), target_item_ids)
self.assertEqual(item_feature[1].shape,
(len(target_item_ids), encoder_size))
output_embedding = {k: v for k, v in zip(*item_feature)}
np.testing.assert_almost_equal(output_embedding[240],
output_embedding[243])
def test_run(self):
# This tests that GraphConvolutionalMatrixCompletion runs without error, and its loss and rmse are small enough.
n_users = 101
n_items = 233
n_data = 3007
am1 = _make_sparse_matrix(n_users, n_items, n_data)
am2 = 2 * _make_sparse_matrix(n_users, n_items, n_data)
adjacency_matrix = am1 + am2
user_ids = adjacency_matrix.tocoo().row
item_ids = adjacency_matrix.tocoo().col
ratings = adjacency_matrix.tocoo().data
dataset = GcmcDataset(user_ids, item_ids, ratings)
graph_dataset = GcmcGraphDataset(dataset, test_size=0.1)
encoder_hidden_size = 100
encoder_size = 100
scope_name = 'GraphConvolutionalMatrixCompletionGraph'
model = GraphConvolutionalMatrixCompletion(
graph_dataset=graph_dataset,
encoder_hidden_size=encoder_hidden_size,
encoder_size=encoder_size,
scope_name=scope_name,
batch_size=1024,
epoch_size=10,
learning_rate=0.01,
dropout_rate=0.7,
normalization_type='symmetric')
reports = model.fit()
test_loss = float(reports[-1].split(',')[-2].split('=')[-1])
test_rmse = float(reports[-1].split(',')[-1].split('=')[-1][:-1])
self.assertLess(test_loss, 1.0)
self.assertLess(test_rmse, 0.7)
def test_with_click_threshold(self):
user_ids = np.array([1, 1, 2, 2, 2, 3])
item_ids = np.array([1, 2, 1, 2, 3, 1])
ratings = np.array([1, 0, 1, 0, 1, 0])
test_size = 0.0
user_features = [{
1: np.array([10, 11]),
2: np.array([20, 21]),
3: np.array([30, 31])
}]
item_features = [{
1: np.array([10, 11, 12]),
2: np.array([20, 21, 22]),
3: np.array([30, 31, 32])
}]
dataset = GcmcDataset(user_ids=user_ids,
item_ids=item_ids,
ratings=ratings,
user_features=user_features,
item_features=item_features)
graph_dataset = GcmcGraphDataset(dataset,
test_size,
min_user_click_count=3)
np.testing.assert_almost_equal([0, 0, 1, 1, 1, 0],
graph_dataset._user.indices)
np.testing.assert_almost_equal([1, 2, 1, 2, 3, 1],
graph_dataset._item.indices)
data = graph_dataset.train_data()
self.assertEqual(item_ids.shape, graph_dataset._item.indices.shape)
self.assertEqual((ratings.shape[0], 2), data['label'].shape)
self.assertEqual(ratings.shape, data['rating'].shape)
self.assertEqual(user_ids.shape, data['user_feature_indices'].shape)
self.assertEqual(item_ids.shape, data['item_feature_indices'].shape)
def test_item_cold_start(self):
n_users = 101
n_items = 233
n_data = 3007
am1 = _make_sparse_matrix(n_users, n_items, n_data)
am2 = 2 * _make_sparse_matrix(n_users, n_items, n_data)
adjacency_matrix = am1 + am2
user_ids = adjacency_matrix.tocoo().row
item_ids = adjacency_matrix.tocoo().col
ratings = adjacency_matrix.tocoo().data
item_features = [{i: np.array([i]) for i in range(n_items)}]
dataset = GcmcDataset(user_ids,
item_ids,
ratings,
item_features=item_features)
graph_dataset = GcmcGraphDataset(dataset, test_size=0.1)
encoder_hidden_size = 100
encoder_size = 100
scope_name = 'GraphConvolutionalMatrixCompletionGraph'
model = GraphConvolutionalMatrixCompletion(
graph_dataset=graph_dataset,
encoder_hidden_size=encoder_hidden_size,
encoder_size=encoder_size,
scope_name=scope_name,
batch_size=1024,
epoch_size=10,
learning_rate=0.01,
dropout_rate=0.7,
normalization_type='symmetric')
model.fit()
user_ids = [90, 62]
item_ids = [11, 236] # 236 is new items
additional_dataset = GcmcDataset(np.array(user_ids),
np.array(item_ids),
np.array([1, 2]),
item_features=[{
236: np.array([236])
}])
results = model.predict_with_new_items(
user_ids, item_ids, additional_dataset=additional_dataset)
self.assertEqual(2, len(results))
self.assertIsNotNone(results[0])
self.assertIsNotNone(results[1])
def test_without_information(self):
user_ids = np.array([1, 1, 2, 2, 2])
item_ids = np.array([1, 2, 1, 2, 3])
ratings = np.array([1, 0, 1, 0, 1])
rating_data = GcmcDataset(user_ids=user_ids,
item_ids=item_ids,
ratings=ratings)
test_size = 0.0
dataset = GcmcGraphDataset(rating_data, test_size)
data = dataset.train_data()
self.assertEqual(user_ids.shape, data['user'].shape)
self.assertEqual(item_ids.shape, data['item'].shape)
self.assertEqual((ratings.shape[0], 2), data['label'].shape)
self.assertEqual(ratings.shape, data['rating'].shape)
self.assertEqual(user_ids.shape, data['user_feature_indices'].shape)
self.assertEqual(item_ids.shape, data['item_feature_indices'].shape)
def test(self):
# This tests that GraphConvolutionalMatrixCompletion runs without error, and its loss and rmse are small enough.
n_users = 101
n_items = 233
n_data = 3007
am1 = _make_sparse_matrix(n_users, n_items, n_data)
am2 = 2 * _make_sparse_matrix(n_users, n_items, n_data)
adjacency_matrix = am1 + am2
user_ids = adjacency_matrix.tocoo().row
item_ids = adjacency_matrix.tocoo().col
ratings = adjacency_matrix.tocoo().data
item_features = [{i: np.array([i]) for i in range(n_items)}]
rating_data = GcmcDataset(user_ids,
item_ids,
ratings,
item_features=item_features)
dataset = GcmcGraphDataset(dataset=rating_data, test_size=0.2)
self.assertEqual(
(n_items + 1, 1),
dataset.item_features[0].shape) # because of default index.
def run(self):
tf.reset_default_graph()
df = self.load_data_frame('train_data', required_columns={self.user_column_name, self.item_column_name, self.rating_column_name})
user_features = self.load('user_features')
item_features = self.load('item_features')
df.drop_duplicates(subset=[self.user_column_name, self.item_column_name], inplace=True)
df = sklearn.utils.shuffle(df)
df = df.head(n=int(self.max_data_size))
user_ids = df[self.user_column_name].values
item_ids = df[self.item_column_name].values
ratings = df[self.rating_column_name].values
dataset = GcmcDataset(user_ids=user_ids, item_ids=item_ids, ratings=ratings, user_features=user_features, item_features=item_features)
graph_dataset = GcmcGraphDataset(dataset=dataset,
test_size=self.test_size,
min_user_click_count=self.min_user_click_count,
max_user_click_count=self.max_user_click_count)
model = GraphConvolutionalMatrixCompletion(graph_dataset=graph_dataset, **self.model_kwargs)
self.task_log['report'] = [str(self.model_kwargs)] + model.fit(try_count=self.try_count, decay_speed=self.decay_speed)
self.dump(self.task_log['report'], 'report')
self.dump(model, 'model')
def test_get_user_feature_with_new_items(self, dummy_get_user_feature):
n_users = 10
n_items = 20
n_data = 3007
am1 = _make_sparse_matrix(n_users, n_items, n_data)
am2 = 2 * _make_sparse_matrix(n_users, n_items, n_data)
adjacency_matrix = am1 + am2
user_ids = adjacency_matrix.tocoo().row
item_ids = adjacency_matrix.tocoo().col
ratings = adjacency_matrix.tocoo().data
item_features = [{i: np.array([i]) for i in range(n_items)}]
dataset = GcmcDataset(user_ids,
item_ids,
ratings,
item_features=item_features)
graph_dataset = GcmcGraphDataset(dataset, test_size=0.1)
encoder_hidden_size = 100
encoder_size = 100
scope_name = 'GraphConvolutionalMatrixCompletionGraph'
model = GraphConvolutionalMatrixCompletion(
graph_dataset=graph_dataset,
encoder_hidden_size=encoder_hidden_size,
encoder_size=encoder_size,
scope_name=scope_name,
batch_size=1024,
epoch_size=10,
learning_rate=0.01,
dropout_rate=0.7,
normalization_type='symmetric')
n_user_embed_dimension = 50
dummy_get_user_feature.return_value = np.zeros(
(len(user_ids) * len(item_ids), n_user_embed_dimension))
user_features = model.get_user_feature_with_new_items(
item_ids, additional_dataset=dataset, with_user_embedding=False)
self.assertEqual(len(user_features[0]), n_users)
self.assertEqual(user_features[1].shape,
(n_users, n_user_embed_dimension))
def test_with_information(self):
user_ids = np.array([1, 1, 2, 2, 2])
item_ids = np.array([1, 2, 1, 2, 3])
ratings = np.array([1, 0, 1, 0, 1])
test_size = 0.0
user_features = [{1: np.array([10, 11]), 2: np.array([20, 21])}]
item_features = [{
1: np.array([10, 11, 12]),
2: np.array([20, 21, 22]),
3: np.array([30, 31, 32])
}]
dataset = GcmcDataset(user_ids=user_ids,
item_ids=item_ids,
ratings=ratings,
user_features=user_features,
item_features=item_features)
graph_dataset = GcmcGraphDataset(dataset, test_size)
data = graph_dataset.train_data()
self.assertEqual(user_ids.shape, data['user'].shape)
self.assertEqual(item_ids.shape, data['item'].shape)
self.assertEqual((ratings.shape[0], 2), data['label'].shape)
self.assertEqual(ratings.shape, data['rating'].shape)
self.assertEqual(user_ids.shape, data['user_feature_indices'].shape)
self.assertEqual(item_ids.shape, data['item_feature_indices'].shape)
def setUp(self) -> None:
dataset = GcmcDataset(user_ids=np.array([0, 1, 2]), item_ids=np.array([10, 11, 12]), ratings=np.array([100, 101, 102]))
self.graph_dataset = GcmcGraphDataset(dataset=dataset, test_size=0.1)
self.additional_dataset = GcmcDataset(user_ids=np.array([1, 2, 3]), item_ids=np.array([13, 14, 15]), ratings=np.array([103, 101, 102]))