def test_histogram_vectorizer_outlier_bins():
vectorizer = HistogramVectorizer(n_components=20, append_outlier_bins=True)
result = vectorizer.fit_transform(value_sequence_data)
assert result.shape == (len(value_sequence_data), 20 + 2)
transform_result = vectorizer.transform([[-1.0, -1.0, -1.0, 150.0]])
assert transform_result[0][0] == 3.0
assert transform_result[0][-1] == 1.0
def test_wass1d_transfomer():
vectorizer = HistogramVectorizer()
histogram_data = vectorizer.fit_transform(value_sequence_data)
transformer = Wasserstein1DHistogramTransformer()
result = transformer.fit_transform(histogram_data)
for i in range(result.shape[0]):
for j in range(i + 1, result.shape[0]):
assert np.isclose(
kantorovich1d(histogram_data[i], histogram_data[j]),
np.sum(np.abs(result[i] - result[j])),
)
def test_histogram_vectorizer_basic():
vectorizer = HistogramVectorizer(n_components=20)
result = vectorizer.fit_transform(value_sequence_data)
assert result.shape == (len(value_sequence_data), 20)
transform_result = vectorizer.transform(value_sequence_data)
assert np.all(result == transform_result)