Installation#

milia needs Python 3.12 or later. The package is published on PyPI as pundamilia and imported as milia.

pip install pundamilia              # core: detectors, clusterers, validation, retrieval
pip install 'pundamilia[all]'       # every extra below

Extras#

The core install has no heavy dependencies. Each extra adds one family of methods, and a method that needs an extra says which one when you call it.

Extra

Adds

plots

figures in lets-plot and matplotlib, and Great Tables

ts

DTW and soft-DTW k-means, k-Shape (tslearn)

tuning

Optuna pipeline search, UMAP and PaCMAP

tensor

Tucker embeddings (tensorly)

foundation

Chronos-2 sequence embeddings

changepoint

PELT and binary segmentation (ruptures)

explain

SHAP explanations of Isolation Forest flags

bench

Diebold-Mariano and Model Confidence Set comparisons

Check the install#

import milia

print(milia.__version__)
print(milia.available("detector"))