Practical Predictive Analytics

Cross-tabulations

Cross-tabulations are a good starting point for examining the relationships between nominal variables. There are many ways to do this in R. I like to start off by using the CrossTable() function from the gmodels package, since it will give not only the cell counts but also the frequencies for the rows and columns totals, and supply a chi-square statistic to measure the statistical significance. It may not be pretty but it does the job well.

See the following code:

install.packages("gmodels") library(gmodels) CrossTable(mtcars$cyl, mtcars$gear, prop.t=TRUE, prop.r=TRUE, prop.c=TRUE) > CrossTable(mtcars$cyl, mtcars$gear, prop.t=TRUE, prop.r=TRUE, prop.c=TRUE)