Derives prediction rule ensembles (PREs). Largely follows the procedure for deriving PREs as described in Friedman & Popescu (2008; <doi:10.1214/07-AOAS148>), with several adjustments and improvements. The main function pre() derives a prediction rule ensemble. Functions coef() and importance() can be used to inspect the generated ensemble. Function predict() generates predicted values. Singeplot() and pairplot() depict dependence of the output on specified predictor variables. Function cvpre() performs full cross validation of a pre to calculate the expected prediction error. Functions interact() and bsnullinteract() can be used to assess interaction effects of predictor variables.
| Version: | 0.1 |
| Depends: | R (≥ 3.3.0), partykit, glmnet, MatrixModels, Formula, methods, graphics |
| Suggests: | akima |
| Published: | 2016-12-24 |
| Author: | Marjolein Fokkema [aut, cre] |
| Maintainer: | Marjolein Fokkema <m.fokkema at fsw.leidenuniv.nl> |
| License: | GPL-2 | GPL-3 |
| NeedsCompilation: | no |
| CRAN checks: | pre results |
| Reference manual: | pre.pdf |
| Package source: | pre_0.1.tar.gz |
| Windows binaries: | r-devel: pre_0.1.zip, r-release: pre_0.1.zip, r-oldrel: not available |
| OS X Mavericks binaries: | r-release: pre_0.1.tgz, r-oldrel: not available |
Please use the canonical form https://CRAN.R-project.org/package=pre to link to this page.