catch22 is a subset of the 22 best-performing time-series features, distilled from a comprehensive library of over 7000 time-series features in hctsa. The purpose of this repository is to guide new users through the practical utilisation of catch22 within their Python-based time-series analysis workflows.
To get started, you'll need to install the latest version of pycatch22 via pip:
pip install pycatch22The catch22 usage examples are organised as individual Jupyter notebooks, each focusing on a different application.
-
catch22_pandas_example.ipynb: This notebook serves as an example for implementing catch22 and constructing a time-series
$\times$ feature matrix in pandas. Here, you will learn how to extract time-series features from a general dataset and integrate them into a pandas data frame, providing a well-organised and accessible format for further time-series analysis.
For a more in-depth understanding of the time-series features comprising catch22, a clear and intuitive explanation of each feature can be found in the catch22 GitBook, including relevant examples and plots. A high-level summary of the time-series feature names, conceptual groupings, and brief descriptions are also provided in a feature overview table.
While this repository specifically focuses on the Python (pycatch22) implementation, catch22 is available in multiple languages. Further details about the Julia (Catch22.jl), R (Rcatch22) and MATLAB implementations of catch22 can be found in the original repository and wiki.
- catch22 - CAnonical Time-series CHaracteristics
- Lubba et al. (2019). catch22: CAnonical Time-series CHaracteristics.
- Add a more general time-series dataset format (i.e. csv format).