Variable Importance

The function relimp uses Dominance Analysis (DA) to rank order the variables in a linear or logistic regression in terms of relative importance.

relimp(x)

Arguments

x

an object of class lm or glm.

Value

a ggplot2 graph

Details

The function is a wrapper for the dominanceAnalysis function in the dominanceanalyis package. For linear regression DA analysis uses change in R-squared to assess the average contribution of each variable to the set all 1, 2, 3, ..., p variable models. For logistic regression, Estrella's Pseudo R-squared is used (see references).

Results are graphed as horizontal dot plots.

References

Estrella, A. (1998). A new measure of fit for equations with dichotomous dependent variables. Journal of Business & Economic Statistics, 16(2), 198-205.

Azen, R., & Budescu, D. V. (2003). The dominance analysis approach for comparing predictors in multiple regression. Psychological Methods, 8(2), 129-148.

Azen, R., & Traxel, N. (2009). Using Dominance Analysis to Determine Predictor Importance in Logistic Regression. Journal of Educational and Behavioral Statistics, 34(3), 319-347.

Examples

data(mtcars)

#######################
# Multiple regression #
#######################
fit <- lm(mpg ~ ., data = auto_mpg)
relimp(fit)
#> working ...


#######################
# Logistic regression #
#######################
fit2 <- glm(caesarian ~ ., family = binomial, data = caesarian)
relimp(fit2)
#> working ...