Giotto-TDA is a well-tested suite of computational topology tools, compatible with the scikit-learn API and framework.
From the docs:
giotto-tda is a high performance topological machine learning toolbox in Python built on top of scikit-learn and is distributed under the GNU AGPLv3 license. It is part of the Giotto family of open-source projects.
Supported data types include:
- point clouds,
- tabular data,
- time series data
- (directed) graphs,
- images.
Supported filtrations include:
- (weighted/sparse) Vietoris-Rips,
- weak alpha filtration,
- euclidean Čech filtration,
- filtered cubical complexes,
- (un)directed flag complex filtrations.
Persistence diagrams can also be converted into other representations including persistence landscapes, persistence images and Betti curves.
In line with the scikit-learn framework, preprocessing, persistent homology and diagram representations can be combined into a single pipeline.
Under the hood, many demanding workloads are implemented in C++, vectorised and parallelised.