This is the code to a paper: https://www.mdpi.com/1099-4300/25/12/1671
If you just wanna use a machine-learning-based estimation of the Hurst exponent, then the "example"-folder will provide you with a working implementation for three test datasets that can easily be adapted to your time series data. Ignore the following.
This repository contains everything about the porposed method to estimate the Hurst exponent from a given time series data using several ML Algorithms.
The code still needs to be cleaned, but except for the financial data sets, which we cannot share, is everything available.
You can downnload the training data from my drive:
https://drive.google.com/drive/folders/1vYdwCgE9l-afB_mm6vpprZ9RWBe7at8z?usp=sharing
The fractional Lévy motion code is from: https://github.com/cpgr/flm/tree/master