FactoMineR (version 1.02)

FDA: Factorial Discriminant Analysis

Description

Performs Factorial Discriminant Analysis (FDA).

Usage

FDA(X, fact, new.data = NULL, new.fact = NULL, prior = NULL, 
    cross.val = FALSE, graph = TRUE)

Arguments

X

a data frame with n rows (individuals) and p columns (including one factor).

fact

a factor specifying the class for each observation.

new.data

a data frame of individuals to be classified or, if new.fact is not null, which formed the test sample.

new.fact

if new.dat is not null, a factor specifying the class for each observation of the test sample.

prior

a vector. The prior probabilities of class membership. If unspecified, the class proportions for the training sample (new.dat) are used.

cross.val

if TRUE, returns results for leave-one-out cross-validation.

graph

boolean, if TRUE graphs are plotted

Value

• Returns a list including :
• eig
  a numeric vector containing all the eigenvalues
• eigen.vectors
  a list of matrices containing all the eigenvectors
• var
  a list of matrices containing all the results for the active variables
• cg
  a list of matrices containing all the results for the centers of gravity
• ind
  a list of matrices containing all the results for the individuals
• call
  a list with the data frame and grouping factor used
• df
  the discriminant functions
• score
  a vector whith the individuals scores.
• eval
  a list with all the results for the evaluation of the affectation model
• res.cv
  a list with all the results for for leave-one-out cross-validation.

Details

If there is no test sample, the evaluation of the affectation model is realised on the test sample itself.

See Also

plot.FDA, print.FDA

Examples

data(wine)
res.fda <- FDA(wine[,-(1:28)], fact=wine[,1])