Research indexPreprint
stream-learn -- open-source Python library for difficult data stream\n batch analysis
Ksieniewicz, Zyblewski
arXiv (Cornell University)29 Jan 2020Open access
NobleIDark:/48914/w1/20260525/75EEFA0BDOI10.48550/arxiv.2001.11077
stream-learn -- open-source Python library for difficult data stream\n batch analysis
Ksieniewicz, Zyblewski · arXiv (Cornell University) · 2020
NobleIDark:/48914/w1/20260525/75EEFA0BDOI10.48550/arxiv.2001.11077
Abstract
stream-learn is a Python package compatible with scikit-learn and developed\nfor the drifting and imbalanced data stream analysis. Its main component is a\nstream generator, which allows to produce a synthetic data stream that may\nincorporate each of the three main concept drift types (i.e. sudden, gradual\nand incremental drift) in their recurring or non-recurring versions. The\npackage allows conducting experiments following established evaluation\nmethodologies (i.e. Test-Then-Train and Prequential). In addition, estimators\nadapted for data stream classification have been implemented, including both\nsimple classifiers and state-of-art chunk-based and online classifier\nensembles. To improve computational efficiency, package utilises its own\nimplementations of prediction metrics for imbalanced binary classification\ntasks.\n