ClipSigma

class sdm.processing.numerical.ClipSigma(*, threshold: float = 4.0, method: Literal['hard', 'log'] = 'hard')

Bases: Processor

Two-stage z-score outlier clipping.

The first pass masks values outside the initial z-score bounds, then the second pass refits bounds on the remaining values. NaN and infinite values are ignored when fitting statistics, and NaN values are preserved.

Parameters:

• threshold (float) – Positive z-score multiplier setting how many standard deviations from the mean mark the clipping bounds.

• method (Literal['hard', 'log']) – "hard" clamps to the fitted bounds; "log" expands them outward by log1p(abs(x)) before clamping each value.

Capabilities

Handled Semantic Types	numerical
Requires Fitting	✅