def test_sample_weight():
model = LightFM()
train = sp.coo_matrix(np.array([[0, 1], [0, 1]]))
with pytest.raises(ValueError):
# Wrong number of weights
sample_weight = sp.coo_matrix(np.zeros((2, 2)))
model.fit(train, sample_weight=sample_weight)
with pytest.raises(ValueError):
# Wrong shape
sample_weight = sp.coo_matrix(np.zeros(2))
model.fit(train, sample_weight=sample_weight)
with pytest.raises(ValueError):
# Wrong order of entries
model.fit(train, sample_weight=sample_weight)
sample_weight = sp.coo_matrix((train.data, (train.row, train.col)))
model.fit(train, sample_weight=sample_weight)
model = LightFM(loss='warp-kos')
with pytest.raises(NotImplementedError):
model.fit(train, sample_weight=np.ones(1))
def test_nan_interactions():
no_users, no_items = 1000, 1000
train = sp.rand(no_users, no_items, format='csr', random_state=42)
train.data *= np.nan
model = LightFM(loss='warp')
with pytest.raises(ValueError):
model.fit(train)
def test_nan_features():
no_users, no_items = 1000, 1000
train = sp.rand(no_users, no_items, format='csr', random_state=42)
features = sp.identity(no_items)
features.data *= np.nan
model = LightFM(loss='warp')
with pytest.raises(ValueError):
model.fit(train,
epochs=10,
user_features=features,
item_features=features)
def test_return_self():
no_users, no_items = 10, 100
train = sp.coo_matrix((no_users, no_items), dtype=np.int32)
model = LightFM()
assert model.fit_partial(train) is model
assert model.fit(train) is model
def test_coo_with_duplicate_entries():
# Calling .tocsr on a COO matrix with duplicate entries
# changes its data arrays in-place, leading to out-of-bounds
# array accesses in the WARP code.
# Reported in https://github.com/lyst/lightfm/issues/117.
rows, cols = 1000, 100
mat = sp.random(rows, cols)
mat.data[:] = 1
# Duplicate entries in the COO matrix
mat.data = np.concatenate((mat.data, mat.data[:1000]))
mat.row = np.concatenate((mat.row, mat.row[:1000]))
mat.col = np.concatenate((mat.col, mat.col[:1000]))
for loss in ('warp', 'bpr', 'warp-kos'):
model = LightFM(loss=loss)
model.fit(mat)
def test_regression_full_batch_predict():
no_components = 2
np.random.seed(42)
ds = RandomDataset(no_items=5, density=1)
model = LightFM(no_components=no_components)
model.fit(ds.train,
user_features=ds.user_features,
item_features=ds.item_features)
# Set non zero biases
model.item_biases += 0.2
model.user_biases += 0.5
user_ids = [0, 1, 2]
model.batch_setup(item_chunks={0: ds.item_ids},
item_features=ds.item_features,
user_features=ds.user_features)
recoms = model.batch_predict(
user_ids=user_ids,
chunk_id=0,
top_k=0, # Score all items
)
zeros = 0
for user_id in user_ids:
scores = model.predict(
user_ids=user_id,
item_ids=ds.item_ids,
item_features=ds.item_features,
user_features=ds.user_features,
num_threads=1,
)
if sum(scores) != 0:
zeros += 1
assert_array_almost_equal(recoms[user_id][1], scores)
assert zeros != 0
def test_exception_on_divergence():
no_users, no_items = 1000, 1000
train = sp.rand(no_users, no_items, format='csr', random_state=42)
model = LightFM(learning_rate=10000000.0, loss='warp')
with pytest.raises(ValueError):
model.fit(train, epochs=10)