RayDistributor for using Ray to distribute the calculations in tsfresh

[TheaperDeng]

Ray is getting popular for building distributed applications and easy to fit into tsfresh by a RayDistributor.

Distributed tsfresh on Ray

This repo involves a new RayDistributor for tsfresh to use ray to distribute the calculations.

RayDistributor is a subclass of IterableDistributorBaseClass in tsfresh which follows the developing instruction in https://tsfresh.readthedocs.io/en/latest/text/tsfresh_on_a_cluster.html.

Quick Start

Use RayDistributor the same way as MultiprocessingDistributor, ClusterDaskDistributor or LocalDaskDistributor.

from tsfresh.utilities.distribution import RayDistributor

distributor = RayDistributor(n_workers=4)
# ...
extracted_features = extract_features(..., distributor=distributor)
# ...

Code change summary

• add RayDistributor definition in tsfresh.utilities.distribution
• add RayDistributor document in docs/text/tsfresh_on_a_cluster.rst
• Update pre-commit-config to enable future development
• Update test-requirements.txt for UT
• munually test the UT and document generation locally

[TheaperDeng]

@nils-braun It would be great to have some suggestions to avoid changing pre-commit-config version and to the PR itself