https://github.com/maxg87/chatterjee-correlations
One-off script to explore Chatterjee Correlation
https://github.com/maxg87/chatterjee-correlations
Last synced: about 1 year ago
JSON representation
One-off script to explore Chatterjee Correlation
- Host: GitHub
- URL: https://github.com/maxg87/chatterjee-correlations
- Owner: MaxG87
- License: gpl-3.0
- Created: 2024-05-07T12:41:57.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2024-05-07T12:52:45.000Z (about 2 years ago)
- Last Synced: 2025-03-15T03:14:02.206Z (over 1 year ago)
- Language: Python
- Size: 19.5 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# chatterjee-correlations
One-off script to explore Chatterjee Correlation
This repository hosts a simple Python file that implements the [Chatterjee Correlation algorithm](https://arxiv.org/abs/1909.10140). Its main goal is to have fun with the implementation and to explore how the claims hold in practice. By no means it is intended to be reusable or a demonstration of good programming practices.
If you are interested in a library to be used, you may find [xicorpy](https://pypi.org/project/xicorpy/) interesting. I haven't used it myself, but it looks promising. Additionally, [this paper](https://arxiv.org/abs/2108.06828) seems to improve the power of the coefficient calculation.
## Results
The results are stunning. For a broad range of nonlinear functions a very strong correlation is detected. This makes this correlation algorithm much more suited to detect hidden relations than my prior go-to method Support Vector Machines with powerful kernels.