An implementation of the replacement bootstrap algorithm for dependent data with modifications to support real-numbered series (up to 64 bits) and accelerations to improve compute performance.
If you use this code or a modification of this code, please reference:
Sani, Amir, Alessandro Lazaric, and Daniil Ryabko. "The replacement bootstrap for dependent data." 2015 IEEE International Symposium on Information Theory (ISIT). IEEE, 2015.
Here's the BibTex:
@inproceedings{sani2015replacement,
title={The replacement bootstrap for dependent data},
author={Sani, Amir and Lazaric, Alessandro and Ryabko, Daniil},
booktitle={2015 IEEE International Symposium on Information Theory (ISIT)},
pages={1194--1198},
year={2015},
organization={IEEE}
}
series - Real or integer series
bootstraps - Number of Bootstraps
seed - Random number seed
replacement_percentage - The percentage of the original series to replace max(1, 0.01 <= replacement_percentage).
replace - Sample with or without replacement [True, False]
rho - Concentration control. rho closer to 0 concentrates bootstraps towards the mean, while rho closer to 1 maximizes variability.
See notebook example in example notebook folder.
Copyright Amir Sani 2016