lmap (version 0.2.4)

mrrr: Multinomial Reduced Rank Regression

Description

The function mrrr performs multinomial reduced rank regression for a nominal response variable and a set of predictor variables.

Usage

mrrr(y, X, S = 2, trace = FALSE, maxiter = 65536, dcrit = 1e-06, start = NULL)

Value

Xoriginal Matrix X from input

X Scaled X matrix

G class indicator matrix

ynames class names of response classes

xnames variable names of the predictors

mx means of the predictor variables

sdx standard deviations of the predictor variables

U coordinate matrix of row objects

B matrix with regression coefficients

V Class coordinate matrix

iters number of iterations

deviance value of the deviance at convergence

Arguments

y

An N vector of the responses (categorical).

X

An N by P matrix with predictor variables

S

Positive number indicating the dimensionality of teh solution

trace

Boolean indicating whether a trace of the algorithm should be printed on the console.

maxiter

maximum number of iterations

dcrit

convergence criterion

start

start values. If start=NULL, the algorithm computes the start values.

Examples

if (FALSE) {
data(dataExample_mru)
y = as.matrix(dataExample_mru[ , 1])
X = as.matrix(dataExample_mru[ , 2:6])
output = mrrr(y = y, X = X, S = 2)
}