A data.frame processor/conditioner that prepares real-world data for predictive modeling in a statistically sound manner. Prepares variables so that data has fewer exceptional cases, making it easier to safely use models in production. Common problems 'vtreat' defends against: Inf, NA, too many categorical levels, rare categorical levels, new categorical levels (levels seen during application, but not during training). 'vtreat::prepare' should be used as you would use 'model.matrix'.
| Version: | 0.5.28 |
| Suggests: | testthat, knitr, parallel, rmarkdown, dplyr, ggplot2 |
| Published: | 2016-10-25 |
| Author: | John Mount, Nina Zumel |
| Maintainer: | John Mount <jmount at win-vector.com> |
| BugReports: | https://github.com/WinVector/vtreat/issues |
| License: | GPL-3 |
| URL: | https://github.com/WinVector/vtreat |
| NeedsCompilation: | no |
| Materials: | README |
| CRAN checks: | vtreat results |
| Reference manual: | vtreat.pdf |
| Vignettes: |
vtreat package vtreat cross frames vtreat grouping example vtreat overfit vtreat Rare Levels vtreat scale mode vtreat significance vtreat data splitting Variable Types |
| Package source: | vtreat_0.5.28.tar.gz |
| Windows binaries: | r-devel: vtreat_0.5.27.zip, r-release: vtreat_0.5.27.zip, r-oldrel: vtreat_0.5.27.zip |
| OS X Mavericks binaries: | r-release: vtreat_0.5.27.tgz, r-oldrel: vtreat_0.5.27.tgz |
| Old sources: | vtreat archive |
Please use the canonical form https://CRAN.R-project.org/package=vtreat to link to this page.