Welcome to wildboar#

Wildboar is a package for supervised and unsupervised machine learning for time series data.

Classification

Identifying which category a time series belong to.

Algorithms: Random shapelet forest, Extra shapelet trees and ROCKET.

Regression

Predicting a continuous-valued attribute associated with a time series.

Algorithms: Random shapelet forest, Extra shapelet trees and ROCKET.

Explainability

Explaining time series classifiers by counterfactual reasoning and by importance scores.

Algorithms: Shapelet forest counterfactuals, KNearest counterfactuals and Prototype counterfactuals.

Unsupervised

Identify interesting regions, motifs or outlier in time series. Or embed time series using various features.

Algorithms: MatrixProfile, Isolation shapelet forests and Shapelet, ROCKET, and interval embeddings.

News

Dec 7, 2022. We have released Version 1.1 with numerous new estimators, functions, fixes, and enhancements.

See the complete list of changes here.

Jun 20, 2022. In anticipation for the Version 1.1, we have released a new release candidate with some fixes and improvements.

See the complete list of changes here.