def test_tf_bmf_batch_accuracy(tf_session):
from lenskit.algorithms import basic
from lenskit.algorithms import bias
import lenskit.crossfold as xf
from lenskit import batch
import lenskit.metrics.predict as pm
ratings = lktu.ml100k.ratings
algo = lktf.BiasedMF(25,
damping=10,
batch_size=1024,
epochs=20,
rng_spec=42)
algo = basic.Fallback(algo, bias.Bias(damping=10))
def eval(train, test):
_log.info('running training')
algo.fit(train)
_log.info('testing %d users', test.user.nunique())
return batch.predict(algo, test)
folds = xf.partition_users(ratings, 5, xf.SampleFrac(0.2))
preds = pd.concat(eval(train, test) for (train, test) in folds)
mae = pm.mae(preds.prediction, preds.rating)
assert mae == approx(0.83, abs=0.025)
user_rmse = preds.groupby('user').apply(
lambda df: pm.rmse(df.prediction, df.rating))
assert user_rmse.mean() == approx(1.03, abs=0.05)
def test_tf_bmf_save_load(tmp_path, tf_session):
"Training, saving, and loading a bias model."
fn = tmp_path / 'bias.bpk'
ratings = lktu.ml_test.ratings
original = lktf.BiasedMF(20, batch_size=1024, epochs=20)
original.fit(ratings)
assert original.user_features_.shape == (ratings.user.nunique(), 20)
assert original.item_features_.shape == (ratings.item.nunique(), 20)
binpickle.dump(original, fn)
_log.info('serialized to %d bytes', fn.stat().st_size)
algo = binpickle.load(fn)
assert algo.bias.mean_ == original.bias.mean_
assert np.all(algo.bias.user_offsets_ == original.bias.user_offsets_)
assert np.all(algo.bias.item_offsets_ == original.bias.item_offsets_)
assert np.all(algo.user_features_ == original.user_features_)
assert np.all(algo.item_features_ == original.item_features_)
preds = algo.predict_for_user(100, [5, 10, 30])
assert all(preds.notna())