js.estimate: A James-Stein Estimator Shrinking to a Uniform Distribution

In bstewart/stm: Estimation of the Structural Topic Model

A James-Stein Estimator Shrinking to a Uniform Distribution

Description

A primarily internal function used in calcfrex.

Usage

js.estimate(prob, ct)

Arguments

prob	the MLE estimate of the discrete probability distribution
ct	the count of words observed to estimate that distribution

Details

This calculates a James-Stein type shrinkage estimator for a discrete probability distribution regularizing towards a uniform distribution. The amount of shrinkage is a function of the variance of MLE and the L2 norm distance from the uniform.

This function is based off the ideas in Hausser and Strimmer (2009)

References

Hausser, Jean, and Korbinian Strimmer. "Entropy inference and the James-Stein estimator, with application to nonlinear gene association networks." Journal of Machine Learning Research 10.Jul (2009): 1469-1484.

bstewart/stm documentation built on Jan. 3, 2024, 6:58 p.m.