Non-Sequential Machine Learning Pipelines with pyWATTS

pyWATTS is an open-source Python-based workflow automation tool for time series analysis. pyWATTS simplifies the evaluation process and the design of repetitive machine learning experiments with time series by providing a powerful pipeline solution including preprocessing and analysis modules. Unlike existing sequential pipeline solutions, pyWATTS enables more complex and non-sequential pipelines. Such non-sequential pipelines are beneficial, for example, in forecasting electrical load time series, detecting anomalies in time series, or generating synthetic time series. This talk presents the basic ideas of pyWATTS, the current features, and existing use cases. It also gives an outlook on the future developments of pyWATTS and the cooperation with sktime.