tsflex: Flexible time series processing & feature extraction

Time series processing and feature extraction are crucial time-intensive steps in conventional machine learning pipelines. Existing packages limited their applicability, as they cannot cope with irregularly-sampled or asynchronous data make strong assumptions about the format. Moreover, these do not focus on execution speed memory efficiency, resulting considerable overhead. We present tsflex , a Python toolkit for time that focuses performance flexibility, enabling broad applicability. This leverages window-stride arguments of same type sequence-index, maintains sequence-index through all operations. is flexible it supports (1) multivariate series, (2) multiple configurations, (3) integrates functions from other packages, while (4) making no sampling regularity, alignment, type. Other functionalities include multiprocessing, detailed logging, chunking sequences, serialization. Benchmarks show faster more memory-efficient compared to similar being permissive its utilization.