FSelect: A Function to perform step-wise discriminant analysis using F statistics

Description

Select data using a F tests

Usage

FSelect(Data, Group, target, p.adj.method = "holm", Missing.Data = "Complete")

Value

FSelect returns list containing at least the following components:

Selected: An ordered list indicating which columns were selected.
F.Selected: An ordered list containing the F statistics for each column indicated in Selected.
PrF: An ordered list containing the p values for each column indicated in Selected.
PrNotes: A string indicating which method was used to control for multiple comparisons
model: An lm object with the final model results.

Arguments

Data: A (non-empty), numeric matrix of data values
Group: A (non-empty), vector indicating group membership. Length(unique(Group))==2
target: The number of desired traits. Target cannot be greater than the number of columns in Data
p.adj.method: The method used to control for false discovery. The default setting is 'holm'
Missing.Data: The method used to handle missing data. The default, 'Complete' will use completeData to impute missing data, setting Missing.Data='Remove' will remove all individuals with missing data. FSelect cannot handle missing data.

References

Costanza M, Afifi A (1979). Comparison of Stopping Rules in Forward Stepwise Discriminant Analysis. Journal of the American Statistical Association, pp. 777 - 78

Habbema J, Hermans J (1977). Selection of Variables in Discriminant Analysis by F - Statistics and Error Rate. Technometrics, 19(4), 487 - 493.

Jennrich R (1977). Stepwise discriminant analysis, volume 3. New York Wiley Sons.

Examples

Run this code

data(Nuclei)
data(Groups)
npcs<-floor(ncol(Nuclei)/5)

dat.comp <- completeData(data = Nuclei, n_pcs = npcs) 
groups.use <- c(1,2) 
use.dat <- which(Groups==groups.use[1]|Groups==groups.use[2])

dat.use <- Nuclei[use.dat,]
GR.use <- Groups[use.dat]

#not run
#FSelect(DAT.use,GR.use,3)