milia.anomaly.models#

Registered anomaly detectors, one module per model (see milia.core.registry).

Modules

controllimit

Shewhart control limit: the largest standardised deviation of any reading from its training mean.

copod

COPOD: a skew-weighted empirical copula.

ecod

ECOD: empirical-CDF tail probabilities, summed over dimensions.

hbos

HBOS: histogram-based outlier score.

iforest

Isolation Forest: how few random splits it takes to isolate a point.

kde

Kernel density estimate: how far into a low-density region a point sits.

knn

k-nearest-neighbour distance to the training data.

lof

Local Outlier Factor: local density relative to a point's own neighbours.

mahalanobis

Robust Mahalanobis distance from a Minimum Covariance Determinant fit.

ocsvm

One-class SVM: distance from a learned support boundary.

reconstruction

Reconstruction error from the leading principal components (the Q statistic of multivariate process control).