Release notes

0.3.1 (2025-06-18)

  • Added tutorial that explains how to interpret the TSProto explanations

  • Fixed issue with incorrect contrastive example plotting when plotted pattern is only in focus class

  • Overriding global_breakpointing parameter, when sbc method is set

  • Bugfix in incorrect number of cluster assignment when sbc method is set

  • Contrastive plots added

  • Bugfix in incorrect barycenters in SBC method

  • UCR ECG dataset example added

0.3.0 (2024-09-11)

  • Added several new clustering models

  • Improved visualizations (added SHAP heatmap)

  • Bugfix with different versions of SHAP inconsistently returning list or 3D numpy array in getshap function

  • Added changepoint_sensitivity instead of pen parameter. Now, the pen parameter is estimated automatically

  • Added support for ROCKET-based clustering with sktime 0.29.0

  • Bugfix with incorrect visualization reported in #9

  • Bugfix with internal data not updating after transform (this is needed for the visualization)

  • Added feature of dynamic cluster determination when float is provided as n_clusters parameter

0.2.0 (2024-04-05)

  • Change the way the cluster number is detected. Now it is detected based on the average number of breakpoints found in the dataset

  • Added parameters that allow to specify SHAP aggregation function for weight calculation

  • Added function that allows to incorporate cluster starts in window as features.

  • Updated documentation

  • Bugixes

0.1.3 (2024-03-05)

  • Modularization of code

  • Scikit-learn interface added

  • Documentation updated

  • Examples added

0.1.0 (2024-02-23)

  • Added pypl installation

  • Added sphinx documentation