BrewPOTS
A set of Jupyter notebook tutorials for PyPOTS, a Python toolbox for modeling partially observed, incomplete time series data.
BrewPOTS is the tutorial companion to PyPOTS, walking through classification, clustering, forecasting and imputation on partially observed time series using the PhysioNet-2012 dataset as a running example. The notebooks are meant to get a PyPOTS user started quickly rather than to reach state-of-the-art results, so model hyperparameters are left untuned; readers are expected to tune them for their own data.
A companion quick-start notebook is also available on Google Colab for readers who want to try it without installing anything locally. PyPOTS itself, which the tutorials teach, has been included as a PyTorch Ecosystem project and is described in an accompanying paper presented at a SIGKDD workshop on time series.
The tutorials are Jupyter notebooks licensed under BSD-3-Clause.