cmdpref: MDPREF model

In cwkwanstat/cmdpref: MDPREF

Description

Metric and non-metric MDPREF model

Usage

cmdpref(pref, ndim = 2, monotone = FALSE, tor = 1e-08, maxit = 50)

Arguments

pref	Preference data matrix. Rows as objects and columns as subjects. Small value for less preferred and large value for more preferred.
ndim	Number of dimensions.
monotone	TRUE for Kruskal monotonic transformation. FALSE for metric scale.
tor	tolerance for monotonic transformation.
maxit	maximum number of interactions for monotonic transformation.

Details

Metric scale:
Singular decomposition is applied to data matrix S.
S = ULV', X = U*sqrt(n-1), Y = V*sqrt(L)/sqrt(n-1), n = no. of objects.

Non-metric scale:
Kruskal monotonic transformation is applied to each subject vector by opscale(level=2,...) in optiscale package.
Metric MDPREF is applied to the transformed data.

Value

An object of class "mdpref".

score	Object coordinates
corr	Subject vectors
d	Singular values
fitted	fitted preference matrix
tpref	transformed preference matrix
stress	Stress I value under monotonic transformation

Author(s)

Chi-wai Kwan

References

Carroll, J. D. (1972). “Individual Differences and Multidimensional Scaling.” In Multidimensional Scaling: Theory and Applications in the Behavioral Sciences, vol. 1, edited by R. N. Shepard, A. K. Romney, and S. B. Nerlove, 105–155. New York: Seminar Press.

Kruskal, Joseph B. (1964) “Nonmetric Multidimensional Scaling: A Numerical Method.” Psychometrika 29: 115-129

Examples

library(lattice)
library(optiscale)
library(cmdpref)
#rank order of preferences of 5 objects for 4 subjects
mydata

fit<-cmdpref(pref=mydata,monotone=TRUE,maxit=100)
summary(fit)

#biplot() arguments are applicable
plot(fit,xlim=c(-2,2),cex=1)

cwkwanstat/cmdpref documentation built on Oct. 23, 2021, 8:21 p.m.