/nn_info

nearest neighbor based estimators for measures of information theory

Primary LanguageJupyter NotebookMIT LicenseMIT

This is a package covering nearest neighbor (nn) based estimators for measures of information theory: entropy, Kullback-Leibler divergence and mutual information. Application is on continous variables that can be multidimensional.

The package includes ideas of several publications (see code), including our contribution: A Bayesian nearest neighbor mutual information estimator (BannMI). BannMI is designed for application on phsophoproteomic data to analyse potential protein-protein interactions.