def test_masked_column_mean(failure_logger, mask_dataset):
mask_value, X_np, X = mask_dataset
mean = _masked_column_mean(X, mask_value).get()
mask = ~sk_get_mask(X_np, value_to_mask=mask_value)
n_columns = X.shape[1]
for i in range(n_columns):
column_mask = mask[:, i]
column_mean = np.mean(X_np[:, i][column_mask])
assert column_mean == mean[i]
def test_masked_column_mode(failure_logger, mask_dataset):
mask_value, X_np, X = mask_dataset
mode = _masked_column_mode(X, mask_value).get()
mask = ~sk_get_mask(X_np, value_to_mask=mask_value)
n_columns = X.shape[1]
for i in range(n_columns):
column_mask = mask[:, i]
column_mode = stats.mode(X_np[:, i][column_mask])[0][0]
assert column_mode == mode[i]
def test_get_mask(failure_logger, mask_dataset):
mask_value, X_np, X = mask_dataset
cu_mask = cu_get_mask(X, value_to_mask=mask_value)
sk_mask = sk_get_mask(X_np, value_to_mask=mask_value)
assert_allclose(cu_mask, sk_mask)