R: Comparison of Relative Importances in a Multinomial Logit...
relrelimp
R Documentation
Comparison of Relative Importances in a
Multinomial Logit Model
Description
Produces a summary
of the relative importance of two predictors or two sets of predictors
in a fitted multinom model object, and compares
relative importances
across two of the fitted logit models.
An index or vector of indices for the effects to be
included in the numerator of the comparison
set2
An index or vector of indices for the effects to be
included in the denominator of the comparison
label1
A character string; mnemonic name for the
variables in set1
label2
A character string; mnemonic name for the
variables in set2
subset
Either a vector of numeric indices for the cases to be included
in the standardization of effects, or a vector of logicals
(TRUE for inclusion)
whose length is the same as the number of rows in the model frame,
object$model.
The default choice is to include all cases in the model frame.
response.cat1
A character
string used to specify the first regression of interest
(i.e., the regression
which predicts the log odds on response.cat1 versus the model's
reference category). The response.cat1 argument should be an
element of object$lab.
response.cat2
A character
string used to specify the second regression of interest
(i.e., the regression
which predicts the log odds on response.cat2 versus the model's
reference category). The response.cat2 argument should be an
element of object$lab.
Details
Computes a relative importance summary as described in
relimp, for each of the two regressions specified by
response.cat1
and response.cat2 (relative to the same
reference category); and computes the
difference of those two relative importance summaries,
along with an estimated
standard error for that difference.
Value
An object of class relrelimp, with at least the following components:
model
The call used to construct the model object summarized
sets
The two sets of indices specified as arguments
response.category
A character vector containing the specified
response.cat1 and response.cat2
log.ratio
The natural logarithm of the ratio of effect
standard deviations corresponding to the two sets specified.
A vector with
three components: the first is for response.cat1
versus the reference
category, the second for response.cat2 versus the
reference category,
the third is the difference.
se.log.ratio
Estimated standard errors for the elements of
log.ratio
## 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.