replacementBootstrap: Entropy-Based Bootstrap For Dependent Data

In damienchallet/RemplacementBootstrap: An Entropy-Based Self-Consistent Bootstrap Procedure

Description

This bootstrap scheme accounts for temporal dependence in a sample by maximizing the relative entropy of the bootstrapped sample with respect to the initial sample. The entropy itself is computed with the lz4 library.

Usage

replacementBootstrap(x,seed=NA,replacementFraction=1.0)

Arguments

x	vector of sample values. It can be an xts object. NAs and Inf values are removed.
seed	random-number generator seed. If specified, set.seed uses this value.
replacementFraction	by default, set to 1. Other values are not implemented for the time being.

Details

The algorithm accounts for serial dependence by maximizing the relative entropy of the bootstrapped samples with respect to that of the original distribution. Entropy values are computed by zipping both sample vectors with lz4 library (see https://github.com/bwlewis/lz4)

Value

A vector of bootstrapped sample values.

Author(s)

Amir Sani and Damien Challet

References

Sani, Amir, Alessandro Lazaric, and Daniil Ryabko. "The replacement bootstrap for dependent data." 2015 IEEE International Symposium on Information Theory (ISIT). IEEE, 2015.

See Also

boot, ~~~

Examples

x=rnorm(100)
replacementBootstrap(x)

damienchallet/RemplacementBootstrap documentation built on May 14, 2019, 3:33 p.m.