conjanal: Conjoint analysis

In cwkwanstat/conjanal: Conjoint analysis

Description

Metric and monotonic conjoint analysis for categorical factors

Usage

conjanal(resp, profile, monotone = F, tor = 1e-10, maxit = 50)

Arguments

resp	Response matrix where rows are profiles and columnes are responses from respondents. Small value for less preferable and large value for more preferable.
profile	Profile matrix where rows are profiles and columens are factors.
monotone	TRUE for Kruskal monotonic transformation. FALSE for metric scale.
tor	Tolerane for monotonic transformation.
maxit	Maximum number of itneractions for monotonic transformation.

Details

Metric conjoint analysis:
Linear regression model

Monotonic conjoint analysis:
Kruskal monotonic transformation is applied to each response column by opscale(level=2,...) in optiscale package. Linear regression is applied to the transformed responses.

Value

A conjanal object

lm	lm object. Results of the linear regression models
part	part worth for each level of each factor of each respondent
imp	Relative importance of each factor for each respondent

Author(s)

Chi-wai Kwan

Examples

library(conjanal)
contrasts(mydata$f1)
contrasts(mydata$f2)
contrasts(mydata$f3)

#define response matrix
resp<-as.matrix(mydata[,4:5])

#define profiles
profile<-mydata[,1:3]

fit.conj<-conjanal(resp=resp,profile=profile,monotone = TRUE)
summary(fit.conj$lm)

#part-worth
fit.conj$part

#relative importance
fit.conj$imp

#predicted utilities
cbind(profile,pred=predict(fit.conj$lm,newdata=profile))

#estimation of market shares
maxconj(fit.conj,csimp)
btlconj(fit.conj,csimp)
logitconj(fit.conj,csimp)

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