## Data on housing and satisfaction, from Venables and Ripley
library(MASS)
library(nnet)
data(housing)
house.mult <- multinom(Sat ~ Infl + Type + Cont, weights = Freq,
data = housing)
relrelimp(house.mult, set1 = 2:3, set2 = 7,
label1 = "Influence", label2 = "Contact",
response.cat1 = "Medium", response.cat2 = "High")
## Computes the relative contribution of Influence and Contact in
## each of the two logistic regressions (Med/Low and High/Low), and
## compares those two relative-contribution measures.
Run the code above in your browser using DataLab