def test_slim_simple_predict():
algo = slim.SLIM()
algo.fit(simple_ratings)
res = algo.predict_for_user(1, [7])
assert res is not None
assert len(res) == 1
assert 7 in res.index
assert not np.isnan(res.loc[7])
def test_slim_simple_predict_binary():
algo = slim.SLIM(regularization=(.05, .05), binary=True)
algo.fit(simple_ratings)
res = algo.predict_for_user(1, [7])
assert res is not None
assert len(res) == 1
assert 7 in res.index
assert not np.isnan(res.loc[7])
def test_slim_multiple_predict():
algo = slim.SLIM()
algo.fit(simple_ratings)
res = algo.predict_for_user(1, [6, 7])
assert res is not None
assert len(res) == 2
assert 6 in res.index
assert 7 in res.index
assert res.index[0] == 6
assert res.index[1] == 7
assert not np.isnan(res.loc[7])
def test_slim_train_big():
algo = slim.SLIM()
algo.fit(ml_ratings)
# Diagonal of the coefficient matrix is 0 and there are some values
assert all(algo.coefficients_.diagonal() == 0)
assert all(np.logical_not(np.isnan(algo.coefficients_.data)))
assert len(algo.coefficients_.data) > 0
res = algo.predict_for_user(1, [7])
assert res is not None
assert len(res) == 1
assert 7 in res.index
assert not np.isnan(res.loc[7])
def test_slim_unordered_predict():
algo = slim.SLIM()
algo.fit(simple_ratings)
res = algo.predict_for_user(1, [7, 6, 9])
assert res is not None
assert len(res) == 3
assert 7 in res.index
assert 6 in res.index
assert 9 in res.index
assert res.index[0] == 7
assert res.index[1] == 6
assert res.index[2] == 9
assert not np.isnan(res.loc[7])
def test_slim_train_binary():
algo = slim.SLIM(regularization=(.05, .05), binary=True)
algo.fit(simple_ratings)
assert isinstance(algo.item_index_, pd.Index)
assert isinstance(algo.user_index_, pd.Index)
# Diagonal of the coefficient matrix is 0 and there are some values
assert all(algo.coefficients_.diagonal() == 0)
assert all(np.logical_not(np.isnan(algo.coefficients_.data)))
assert len(algo.coefficients_.data) > 0
# 7 is associated with 9
seven, nine = algo.item_index_.get_indexer([7, 9])
_log.info('seven: %d', seven)
_log.info('nine: %d', nine)
assert algo.coefficients_[seven, nine] > 0
def test_slim_predict_binary_big_parallel():
algo = slim.SLIM(binary=True, nprocs=5)
algo.fit(ml_ratings)
# Diagonal of the coefficient matrix is 0 and there are some values
assert all(algo.coefficients_.diagonal() == 0)
assert all(np.logical_not(np.isnan(algo.coefficients_.data)))
assert len(algo.coefficients_.data) > 0
res = algo.predict_for_user(1, [7])
assert res is not None
assert len(res) == 1
assert 7 in res.index
assert not np.isnan(res.loc[7])
res = algo.predict_for_user(1)
assert res is not None
assert len(res) == len(ml_ratings.item.unique())
def test_slim_train_smoke_test():
algo = slim.SLIM()
algo.fit(simple_ratings)
def test_slim_init_warn_negative_regularization():
try:
algo = slim.SLIM(-1)
except ValueError:
pass # this is fine
def test_slim_init():
algo = slim.SLIM()
def test_fsslim_init_warn_invalid_selector():
try:
algo = slim.fsSLIM(selector=slim.SLIM())
except ValueError:
pass # this is fine
