Forecastable Component Analysis (ForeCA) is a novel dimension reduction (DR) technique for temporally dependent signals. Contrary to other popular DR methods, such as PCA or ICA, ForeCA explicitly searches for the most ”forecastable” signal. The measure of forecastability is based on negative Shannon entropy of the spectral density of the transformed signal. This R package provides the main algorithms and auxiliary function(summary, plotting, etc) to apply ForeCA to multivariate data (time series).
| Version: | 0.1 |
| Depends: | R (≥ 2.15.0), ifultools (≥ 2.0-0), splus2R (≥ 1.2-0), nlme (≥ 3.1-64) |
| Imports: | R.utils, sapa, mgcv, astsa |
| Published: | 2014-03-03 |
| Author: | Georg M. Goerg |
| Maintainer: | Georg M. Goerg <im at gmge.org> |
| License: | GPL-2 |
| URL: | http://www.gmge.org |
| NeedsCompilation: | no |
| Citation: | ForeCA citation info |
| Materials: | NEWS |
| In views: | TimeSeries |
| CRAN checks: | ForeCA results |
| Reference manual: | ForeCA.pdf |
| Package source: | ForeCA_0.1.tar.gz |
| Windows binaries: | r-devel: ForeCA_0.1.zip, r-release: ForeCA_0.1.zip, r-oldrel: ForeCA_0.1.zip |
| OS X Snow Leopard binaries: | r-release: ForeCA_0.1.tgz, r-oldrel: ForeCA_0.1.tgz |
| OS X Mavericks binaries: | r-release: ForeCA_0.1.tgz |
| Old sources: | ForeCA archive |