R package for Binary Local Expert Regression, a new ensemble machine learning methodology for better risk quantification
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
R
man
vignettes
.Rbuildignore
.gitignore
DESCRIPTION
NAMESPACE
README.md
localexpeRt.Rproj

README.md

Github Issues GitHub version

This project was partially funded via the Omar N. Bradley Officer Research Fellowship in Mathematics.

localexpeRt

This R package is for Binary Local Expert Regression, which is a proprietary ensemble method for regression tasks. Briefly, it involves decomposing the task of predicting a continuous target variable into many simpler probabilistic class estimation tasks. These 'local expert' models are reconstructed to form a unique distributional understanding of each separate instance, and a separate regression model is trained on top of this model output (see David Wolpert's work on stacked generalization).

Read the vignette for this project to see the method in action.

This vignette walks through an application of Binary Local Expert Regression to chemical research data from Max Kuhn's Applied Predictive Modeling package, and serves as an introduction to the methodology.

Install the development version from github

If you have the devtools package installed in R you can download the (unstable) development version like this:

devtools::install_github('nnormandin/localexpeRt')