fedeca.metrics

Define metrics for ECA analysis.

standardized_mean_diff(confounders, treated, weights=None, use_unweighted_variance=True)

Compute the Standardized Mean Differences (SMD).

Compute the Standardized Mean Differences between treated and control patients.

Parameters:

• confounders (pd.DataFrame) – confounders array.

• treated (np.ndarray) – mask of booleans giving information about treated patients.

• weights (np.ndarray) – weights for the aggregation

• use_unweighted_variance (bool) – if True, the variance is computed without weights. To follow https://stats.stackexchange.com/questions/618643/formula-for-standardized-mean-difference-in-cobalt-package-for-categorical-varia # noqa: E501 If False use recalibrated variance as in https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4626409/. # noqa: E501

Returns:

smd – standardized mean differences of the confounders.

Return type:

np.ndarray