import numpy as np import pytest from numpy.testing import assert_allclose, assert_array_equal from pytest import approx from sklearn._config import config_context from sklearn.utils._array_api import ( _convert_to_numpy, get_namespace, yield_namespace_device_dtype_combinations, ) from sklearn.utils._array_api import device as array_device from sklearn.utils.estimator_checks import _array_api_for_tests from sklearn.utils.fixes import np_version, parse_version from sklearn.utils.stats import _weighted_percentile @pytest.mark.parametrize("average", [True, False]) @pytest.mark.parametrize("size", [10, 15]) def test_weighted_percentile_matches_median(size, average): """Ensure `_weighted_percentile` matches `median` when expected. With unit `sample_weight`, `_weighted_percentile` should match the median except when `average=False` and the number of samples is even. For an even array and `average=False`, `percentile_rank=50` gives the lower of the two 'middle' values, that are averaged when calculating the `median`. """ y = np.arange(size) sample_weight = np.ones_like(y) score = _weighted_percentile(y, sample_weight, 50, average=average) # `_weighted_percentile(average=False)` does not match `median` when n is even if size % 2 == 0 and average is False: assert score != np.median(y) else: assert approx(score) == np.median(y) @pytest.mark.parametrize("average", [True, False]) @pytest.mark.parametrize("percentile_rank", [20, 35, 61, [5, 47]]) @pytest.mark.parametrize("size", [10, 15]) def test_weighted_percentile_matches_numpy( global_random_seed, size, percentile_rank, average ): """Check `_weighted_percentile` with unit weights is correct. `average=True` results should be the same as `np.percentile`'s 'averaged_inverted_cdf'. `average=False` results should be the same as `np.percentile`'s 'inverted_cdf'. Note `np.percentile` is the same as `np.quantile` except `q` is in range [0, 100]. We parametrize through different `percentile_rank` and `size` to ensure we get cases where `g=0` and `g>0` (see Hyndman and Fan 1996 for details). """ rng = np.random.RandomState(global_random_seed) y = rng.randint(20, size=size) sw = np.ones_like(y) score = _weighted_percentile(y, sw, percentile_rank, average=average) if average: method = "averaged_inverted_cdf" else: method = "inverted_cdf" assert approx(score) == np.percentile(y, percentile_rank, method=method) @pytest.mark.parametrize("percentile_rank", [50, 100]) def test_weighted_percentile_plus_one_clip_max(percentile_rank): """Check `j+1` index is clipped to max, when `average=True`. `percentile_plus_one_indices` can exceed max index when `percentile_indices` is already at max index. Note that when `g` (Hyndman and Fan) / `fraction_above` is greater than 0, `j+1` (Hyndman and Fan) / `percentile_plus_one_indices` is calculated but never used, so it does not matter what this value is. When percentile of percentile rank 100 falls exactly on the last value in the `weighted_cdf`, `g=0` and `percentile_indices` is at max index. In this case we set `percentile_plus_one_indices` to be max index as well, so the result is the average of 2x the max index (i.e. last value of `weighted_cdf`). """ # Note for both `percentile_rank`s 50 and 100,`percentile_indices` is already at # max index y = np.array([[0, 0], [1, 1]]) sw = np.array([[0.1, 0.2], [2, 3]]) score = _weighted_percentile(y, sw, percentile_rank, average=True) for idx in range(2): assert score[idx] == approx(1.0) def test_weighted_percentile_equal(): """Check `weighted_percentile` with unit weights and all 0 values in `array`.""" y = np.zeros(102, dtype=np.float64) sw = np.ones(102, dtype=np.float64) score = _weighted_percentile(y, sw, 50) assert approx(score) == 0 # XXX: is this really what we want? Shouldn't we raise instead? # https://github.com/scikit-learn/scikit-learn/issues/31032 def test_weighted_percentile_all_zero_weights(): """Check `weighted_percentile` with all weights equal to 0 returns last index.""" y = np.arange(10) sw = np.zeros(10) value = _weighted_percentile(y, sw, 50) assert approx(value) == 9.0 @pytest.mark.parametrize("average", [True, False]) @pytest.mark.parametrize("percentile_rank, expected_value", [(0, 2), (50, 3), (100, 5)]) def test_weighted_percentile_ignores_zero_weight( average, percentile_rank, expected_value ): """Check leading, trailing and middle 0 weights behave correctly. Check that leading zero-weight observations are ignored when `percentile_rank=0`. See #20528 for details. Check that when `average=True` and the `j+1` ('plus one') index has sample weight of 0, it is ignored. Also check that trailing zero weight observations are ignored (e.g., when `percentile_rank=100`). """ y = np.array([0, 1, 2, 3, 4, 5, 6]) sw = np.array([0, 0, 1, 1, 0, 1, 0]) value = _weighted_percentile( np.vstack((y, y)).T, np.vstack((sw, sw)).T, percentile_rank, average=average ) for idx in range(2): assert approx(value[idx]) == expected_value @pytest.mark.parametrize("average", [True, False]) @pytest.mark.parametrize("percentile_rank", [20, 35, 50, 61]) def test_weighted_percentile_frequency_weight_semantics( global_random_seed, percentile_rank, average ): """Check integer weights give the same result as repeating values.""" rng = np.random.RandomState(global_random_seed) x = rng.randint(20, size=10) weights = rng.choice(5, size=10) x_repeated = np.repeat(x, weights) percentile_weights = _weighted_percentile( x, weights, percentile_rank, average=average ) percentile_repeated = _weighted_percentile( x_repeated, np.ones_like(x_repeated), percentile_rank, average=average ) assert percentile_weights == approx(percentile_repeated) # Also check `percentile_rank=50` matches `median` if percentile_rank == 50 and average: assert percentile_weights == approx(np.median(x_repeated)) @pytest.mark.parametrize("constant", [5, 8]) @pytest.mark.parametrize("average", [True, False]) @pytest.mark.parametrize("percentile_rank", [20, 35, 50, 61, [20, 35, 50, 61]]) def test_weighted_percentile_constant_multiplier( global_random_seed, percentile_rank, average, constant ): """Check multiplying weights by a constant does not change the result. Note scale invariance does not always hold when multiplying by a float due to cumulative sum numerical error (which grows proportional to n). """ rng = np.random.RandomState(global_random_seed) x = rng.randint(20, size=20) weights = rng.choice(5, size=20) weights_multiplied = weights * constant percentile = _weighted_percentile(x, weights, percentile_rank, average=average) percentile_multiplier = _weighted_percentile( x, weights_multiplied, percentile_rank, average=average ) assert percentile == approx(percentile_multiplier) @pytest.mark.parametrize("percentile_rank", [50, [20, 35, 50]]) @pytest.mark.parametrize("average", [True, False]) def test_weighted_percentile_2d(global_random_seed, percentile_rank, average): """Check `_weighted_percentile` behaviour is correct when `array` is 2D.""" # Check for when array 2D and sample_weight 1D rng = np.random.RandomState(global_random_seed) x1 = rng.randint(10, size=10) w1 = rng.choice(5, size=10) x2 = rng.randint(20, size=10) x_2d = np.vstack((x1, x2)).T wp = _weighted_percentile( x_2d, w1, percentile_rank=percentile_rank, average=average ) if isinstance(percentile_rank, list): p_list = [] for pr in percentile_rank: p_list.append( [ _weighted_percentile( x_2d[:, i], w1, percentile_rank=pr, average=average ) for i in range(x_2d.shape[1]) ] ) p_axis_0 = np.stack(p_list, axis=-1) assert wp.shape == (x_2d.shape[1], len(percentile_rank)) else: # percentile_rank is scalar p_axis_0 = [ _weighted_percentile( x_2d[:, i], w1, percentile_rank=percentile_rank, average=average ) for i in range(x_2d.shape[1]) ] assert wp.shape == (x_2d.shape[1],) assert_allclose(wp, p_axis_0) # Check when array and sample_weight both 2D w2 = rng.choice(5, size=10) w_2d = np.vstack((w1, w2)).T wp = _weighted_percentile( x_2d, w_2d, percentile_rank=percentile_rank, average=average ) if isinstance(percentile_rank, list): p_list = [] for pr in percentile_rank: p_list.append( [ _weighted_percentile( x_2d[:, i], w_2d[:, i], percentile_rank=pr, average=average ) for i in range(x_2d.shape[1]) ] ) p_axis_0 = np.stack(p_list, axis=-1) assert wp.shape == (x_2d.shape[1], len(percentile_rank)) else: # percentile_rank is scalar p_axis_0 = [ _weighted_percentile( x_2d[:, i], w_2d[:, i], percentile_rank=percentile_rank, average=average ) for i in range(x_2d.shape[1]) ] assert wp.shape == (x_2d.shape[1],) assert_allclose(wp, p_axis_0) @pytest.mark.parametrize( "array_namespace, device, dtype_name", yield_namespace_device_dtype_combinations() ) @pytest.mark.parametrize( "data, weights, percentile", [ # NumPy scalars input (handled as 0D arrays on array API) (np.float32(42), np.int32(1), 50), # Random 1D array, constant weights (lambda rng: rng.rand(50), np.ones(50).astype(np.int32), 50), # Random 2D array and random 1D weights (lambda rng: rng.rand(50, 3), lambda rng: rng.rand(50).astype(np.float32), 75), # Random 2D array and random 2D weights ( lambda rng: rng.rand(20, 3), lambda rng: rng.rand(20, 3).astype(np.float32), [25, 75], ), # zero-weights and `rank_percentile=0` (#20528) (`sample_weight` dtype: int64) (np.array([0, 1, 2, 3, 4, 5]), np.array([0, 0, 1, 1, 1, 0]), 0), # np.nan's in data and some zero-weights (`sample_weight` dtype: int64) (np.array([np.nan, np.nan, 0, 3, 4, 5]), np.array([0, 1, 1, 1, 1, 0]), 0), # `sample_weight` dtype: int32 ( np.array([0, 1, 2, 3, 4, 5]), np.array([0, 1, 1, 1, 1, 0], dtype=np.int32), [25, 75], ), ], ) def test_weighted_percentile_array_api_consistency( global_random_seed, array_namespace, device, dtype_name, data, weights, percentile ): """Check `_weighted_percentile` gives consistent results with array API.""" xp = _array_api_for_tests(array_namespace, device) # Skip test for percentile=0 edge case (#20528) on namespace/device where # xp.nextafter is broken. This is the case for torch with MPS device: # https://github.com/pytorch/pytorch/issues/150027 zero = xp.zeros(1, device=device) one = xp.ones(1, device=device) if percentile == 0 and xp.all(xp.nextafter(zero, one) == zero): pytest.xfail(f"xp.nextafter is broken on {device}") rng = np.random.RandomState(global_random_seed) X_np = data(rng) if callable(data) else data weights_np = weights(rng) if callable(weights) else weights # Ensure `data` of correct dtype X_np = X_np.astype(dtype_name) result_np = _weighted_percentile(X_np, weights_np, percentile) # Convert to Array API arrays X_xp = xp.asarray(X_np, device=device) weights_xp = xp.asarray(weights_np, device=device) with config_context(array_api_dispatch=True): result_xp = _weighted_percentile(X_xp, weights_xp, percentile) assert array_device(result_xp) == array_device(X_xp) assert get_namespace(result_xp)[0] == get_namespace(X_xp)[0] result_xp_np = _convert_to_numpy(result_xp, xp=xp) assert result_xp_np.dtype == result_np.dtype assert result_xp_np.shape == result_np.shape assert_allclose(result_np, result_xp_np) # Check dtype correct (`sample_weight` should follow `array`) if dtype_name == "float32": assert result_xp_np.dtype == result_np.dtype == np.float32 else: assert result_xp_np.dtype == np.float64 @pytest.mark.parametrize("average", [True, False]) @pytest.mark.parametrize("sample_weight_ndim", [1, 2]) def test_weighted_percentile_nan_filtered( global_random_seed, sample_weight_ndim, average ): """Test `_weighted_percentile` ignores NaNs. Calling `_weighted_percentile` on an array with nan values returns the same results as calling `_weighted_percentile` on a filtered version of the data. We test both with sample_weight of the same shape as the data and with one-dimensional sample_weight. """ rng = np.random.RandomState(global_random_seed) array_with_nans = rng.rand(100, 10) array_with_nans[rng.rand(*array_with_nans.shape) < 0.5] = np.nan nan_mask = np.isnan(array_with_nans) if sample_weight_ndim == 2: sample_weight = rng.randint(1, 6, size=(100, 10)) else: sample_weight = rng.randint(1, 6, size=(100,)) # Find the weighted percentile on the array with nans: results = _weighted_percentile(array_with_nans, sample_weight, 30, average=average) # Find the weighted percentile on the filtered array: filtered_array = [ array_with_nans[~nan_mask[:, col], col] for col in range(array_with_nans.shape[1]) ] if sample_weight.ndim == 1: sample_weight = np.repeat(sample_weight, array_with_nans.shape[1]).reshape( array_with_nans.shape[0], array_with_nans.shape[1] ) filtered_weights = [ sample_weight[~nan_mask[:, col], col] for col in range(array_with_nans.shape[1]) ] expected_results = np.array( [ _weighted_percentile( filtered_array[col], filtered_weights[col], 30, average=average ) for col in range(array_with_nans.shape[1]) ] ) assert_array_equal(expected_results, results) @pytest.mark.parametrize( "percentile_rank, expected", [ (90, [np.nan, 5]), ([50, 90], [[np.nan, np.nan], [2.0, 5.0]]), ], ) def test_weighted_percentile_all_nan_column(percentile_rank, expected): """Check that nans are ignored in general, except for all NaN columns.""" array = np.array( [ [np.nan, 5], [np.nan, 1], [np.nan, np.nan], [np.nan, np.nan], [np.nan, 2], [np.nan, np.nan], ] ) weights = np.ones_like(array) values = _weighted_percentile(array, weights, percentile_rank) # The percentile of the second column should be `5` even though there are many nan # values present; the percentile of the first column can only be nan, since there # are no other possible values: assert np.array_equal(values, expected, equal_nan=True) @pytest.mark.skipif( np_version < parse_version("2.0"), reason="np.quantile only accepts weights since version 2.0", ) @pytest.mark.parametrize("percentile", [66, 10, 50]) @pytest.mark.parametrize("average", [False, True]) @pytest.mark.parametrize("uniform_weight", [False, True]) def test_weighted_percentile_like_numpy_quantile( percentile, average, uniform_weight, global_random_seed ): """Check `_weighted_percentile` is equivalent to `np.quantile` with weights.""" # TODO: remove the following skip once no longer applicable. if average and not uniform_weight: pytest.skip( "np.quantile does not support weights with method='averaged_inverted_cdf'" ) rng = np.random.RandomState(global_random_seed) array = rng.rand(10, 100) if uniform_weight: sample_weight = np.ones_like(array) * rng.randint(1, 6, size=1) else: sample_weight = rng.randint(1, 6, size=(10, 100)) percentile_weighted_percentile = _weighted_percentile( array, sample_weight, percentile, average=average ) percentile_numpy_quantile = np.quantile( array, percentile / 100, weights=sample_weight if not uniform_weight else None, method="averaged_inverted_cdf" if average else "inverted_cdf", axis=0, ) assert_array_equal(percentile_weighted_percentile, percentile_numpy_quantile) @pytest.mark.skipif( np_version < parse_version("2.0"), reason="np.nanquantile only accepts weights since version 2.0", ) @pytest.mark.parametrize("percentile", [66, 10, 50]) @pytest.mark.parametrize("average", [False, True]) @pytest.mark.parametrize("uniform_weight", [False, True]) def test_weighted_percentile_like_numpy_nanquantile( percentile, average, uniform_weight, global_random_seed ): """Check `_weighted_percentile` equivalent to `np.nanquantile` with weights.""" # TODO: remove the following skip once no longer applicable. if average and not uniform_weight: pytest.skip( "np.nanquantile does not support weights with " "method='averaged_inverted_cdf'" ) rng = np.random.RandomState(global_random_seed) array_with_nans = rng.rand(10, 100) array_with_nans[rng.rand(*array_with_nans.shape) < 0.5] = np.nan if uniform_weight: sample_weight = np.ones_like(array_with_nans) * rng.randint( 1, 6, size=1, ) else: sample_weight = rng.randint(1, 6, size=(10, 100)) percentile_weighted_percentile = _weighted_percentile( array_with_nans, sample_weight, percentile, average=average ) percentile_numpy_nanquantile = np.nanquantile( array_with_nans, percentile / 100, weights=sample_weight if not uniform_weight else None, method="averaged_inverted_cdf" if average else "inverted_cdf", axis=0, ) assert_array_equal(percentile_weighted_percentile, percentile_numpy_nanquantile)