HiDimDA (version 0.1-1)

HiDimDA-package: High Dimensional Discriminant Analysis

Description

Performs Linear Discriminant Analysis in High Dimensional problems based on covariance estimators derived from low dimensional factor models. Includes routines for classifier training, prediction, cross-validation and variable selection.

Details

ll{ Package: HiDimDA Type: Package Version: 0.1-1 Date: 2011-06-13 License: GPL-3 LazyLoad: yes LazyData: yes }

References

Pedro Duarte Silva, A. (2011) Two Group Classification with High-Dimensional Correlated Data: A Factor Model Approach, Computational Statistics and Data Analysis, doi:10.1016/j.csda.2011.05.002.

See Also

RFlda, predict.RFlda, SinghDS

Examples

#train classifier on Singh's Prostate Cancer Data set, 
# selecting genes by the Extended HC scheme 

ldarule <- RFlda(SinghDS[,-1],SinghDS[,1])     

# get in-sample classification results

predict(ldarule,SinghDS[,-1],grpcodes=levels(SinghDS[,1]))$class         	       

# compare classifications with true assignments

cat("Original classes:
")
print(SinghDS[,1])             		 

# show set of selected genes

cat("Genes kept in discrimination rule:
")
print(colnames(SinghDS)[ldarule$vkpt])             		 
cat("Number of selected genes =",ldarule$nvkpt,"")