scorers

Pass the absolute model score through as the anomaly score.

TimeSeries scorer contract, distinct from legacy scored-table components.

Convert model outputs into standardized anomaly scores.

Cauchy negative log likelihood using robust median and half-IQR estimates.

Signed per-component difference ``actual - predicted`` (no fitting).

Exponential negative log likelihood for nonnegative prediction samples.

Gamma negative log likelihood with method-of-moments sample parameters.

Gaussian negative log likelihood using sample mean and standard deviation.

Wrap a scoring callable or estimator with explicit anomaly direction.

Euclidean distance to the closest fitted normal-window cluster center.

Laplace negative log likelihood using sample median and absolute deviation.

Negative log likelihood fitted separately to prediction samples per point.

Absolute component residuals or their vector norm at each timestamp.

Poisson negative log likelihood for nonnegative integer observations.

Adapt a PyOD-compatible estimator's fit/decision_function to windows.

Convert model outputs into standardized anomaly scores.

One-dimensional Wasserstein distance from each window to training values.

Fit and score sliding windows without crossing series boundaries.

Normalize a model score column with a z-score.