def test_digits_cosine_sieve_batch():
return
model = SumRedundancySelection(100,
'cosine',
random_state=0,
reservoir=X_digits)
model.partial_fit(X_digits)
print("[" + ", ".join(map(str, model.ranking)) + "]")
print("[" + ", ".join([str(round(gain, 4)) for gain in model.gains]) + "]")
assert_array_equal(model.ranking, digits_cosine_sieve_ranking)
assert_array_almost_equal(model.gains, digits_cosine_sieve_gains, 4)
assert_array_almost_equal(model.subset, X_digits[model.ranking])
def test_digits_cosine_sieve_minibatch():
return
model = SumRedundancySelection(100,
'cosine',
random_state=0,
reservoir=X_digits)
model.partial_fit(X_digits[:300])
model.partial_fit(X_digits[300:500])
model.partial_fit(X_digits[500:])
assert_array_equal(model.ranking, digits_cosine_sieve_ranking)
assert_array_almost_equal(model.gains, digits_cosine_sieve_gains, 4)
assert_array_almost_equal(model.subset, X_digits[model.ranking])
