weighted_variance_stability.py

import numpy as np
from sklearn.utils.extmath import _incremental_mean_and_var
from sklearn.utils.sparsefuncs import _incr_mean_var_axis0
from sklearn.utils.sparsefuncs import _csc_mean_var_axis0
from sklearn.utils.sparsefuncs import _csr_mean_var_axis0
from scipy.sparse import csr_matrix
from scipy.sparse import csc_matrix

for dtype in [np.float64, np.float32]:
    print(f"## dtype={dtype.__name__}")
    rng = np.random.RandomState(0)
    X = np.full(
        shape=(1000, 1),
        fill_value=100.,
        dtype=dtype,
    )
    # Add some missing records which should be ignored:
    missing_indices = rng.choice(np.arange(X.shape[0]), 10, replace=False)
    X[missing_indices, 0] = np.nan

    # Random positive weights:
    sample_weight = rng.rand(X.shape[0]).astype(dtype)

    mean, var, _ = _incremental_mean_and_var(
        X,
        last_mean=np.zeros(X.shape[1], dtype=dtype),
        last_variance=np.zeros(X.shape[1], dtype=dtype),
        last_sample_count=0,
        sample_weight=sample_weight,
    )
    print(_incremental_mean_and_var.__name__, mean, var)

    mean, var = _csr_mean_var_axis0(
        csr_matrix(X),
        weights=sample_weight,
    )
    print(_csr_mean_var_axis0.__name__, mean, var)

    mean, var, _ = _incr_mean_var_axis0(
        csr_matrix(X),
        last_mean=np.zeros(X.shape[1], dtype=dtype),
        last_var=np.zeros(X.shape[1], dtype=dtype),
        last_n=np.zeros(shape=(1,), dtype=dtype),
        weights=sample_weight,
    )
    print(_incr_mean_var_axis0.__name__, "csr", mean, var)

    mean, var = _csc_mean_var_axis0(
        csc_matrix(X),
        weights=sample_weight,
    )
    print(_csc_mean_var_axis0.__name__, mean, var)

    mean, var, _ = _incr_mean_var_axis0(
        csc_matrix(X),
        last_mean=np.zeros(X.shape[1], dtype=dtype),
        last_var=np.zeros(X.shape[1], dtype=dtype),
        last_n=np.zeros(shape=(1,), dtype=dtype),
        weights=sample_weight,
    )
    print(_incr_mean_var_axis0.__name__, "csc", mean, var)

I have edited the gist to add some nan values and to print the value of the mean.