Supporting functionality to run 'caret' with spatial or spatial-temporal data. 'caret' is a frequently used package for model training and prediction using machine learning. This package includes functions to improve spatial-temporal modelling tasks using 'caret'. It prepares data for Leave-Location-Out and Leave-Time-Out cross-validation which are target-oriented validation strategies for spatial-temporal models. To decrease overfitting and improve model performances, the package implements a forward feature selection that selects suitable predictor variables in view to their contribution to the target-oriented performance.
| Version: | 0.2.1 |
| Depends: | R (≥ 3.1.0) |
| Imports: | caret, stats, utils, ggplot2 |
| Suggests: | doParallel, GSIF, randomForest, lubridate, raster, sp, knitr, mapview, rmarkdown |
| Published: | 2018-07-12 |
| Author: | Hanna Meyer [cre, aut], Chris Reudenbach [ctb], Marvin Ludwig [ctb], Thomas Nauss [ctb] |
| Maintainer: | Hanna Meyer <hanna.meyer at geo.uni-marburg.de> |
| License: | GPL (≥ 3) | file LICENSE |
| URL: | https://github.com/environmentalinformatics-marburg/CAST |
| NeedsCompilation: | no |
| Materials: | README NEWS |
| CRAN checks: | CAST results |
| Reference manual: | CAST.pdf |
| Vignettes: |
Introduction to CAST |
| Package source: | CAST_0.2.1.tar.gz |
| Windows binaries: | r-devel: CAST_0.2.1.zip, r-release: CAST_0.2.1.zip, r-oldrel: CAST_0.2.1.zip |
| OS X binaries: | r-release: CAST_0.2.1.tgz, r-oldrel: CAST_0.2.1.tgz |
| Old sources: | CAST archive |
Please use the canonical form https://CRAN.R-project.org/package=CAST to link to this page.