aeon v1.3.0

See our changelog to view all changes for this release.

Highlights

• A massive update to the forecasting module and its framework. New forecasters include ARIMA/AutoARIMA, SETAR-Tree/Forest, Theta, TAR/SETAR, and deep learners. Thank you to all contributors including our GSoC 2024 students!
• The shapelet transform and extending algorithms now accept unequal length series and have had a massive speed-up
• KNN and KMeans algorithms now have n_jobs support for multithreading
• A download function for the RehabPile dataset collection has been implemented
• An implementation of ESMOTE for imbalanced classification problems has been added to the collection transformers module
• The Time Mixup Contrastive Learning (TimeMCL) algorithm has been added to the self supervised module housed in the transformations package

New Contributors

• @yarikoptic made their first contribution in #2653
• @CodeFor2001 made their first contribution in #2941
• @angus924 made their first contribution in #3003
• @PipaFlores made their first contribution in #3029
• @sarthakuwar made their first contribution in #2993
• @OutragedOctopus made their first contribution in #3036

All Contributors

@alexbanwell1, @angus924, @aryanpola, @baraline, @chrisholder, @CodeFor2001, @hadifawaz1999, @inclinedadarsh, @Kaustbh, @kavya-r30, @LinGinQiu, @lucifer4073, @MatthewMiddlehurst, @OutragedOctopus, @PipaFlores, @Ramana-Raja, @sarthakuwar, @shinymack, @TinaJin0228, @TonyBagnall, @yarikoptic

Full Changelog: v1.2.0...v1.3.0