Practical Predictive Analytics

Binning character data

Character data is usually grouped according to some sort of hierarchy. But occasionally you will want to group it based upon a text pattern contained in the actual string. Here is an example of binning character data based upon the year (the first four characters of cats):

cats <- as.factor(c('2016-1','2016-2','2016-3')) sales <- c(10,20,30) x <- cbind.data.frame(cats,sales) x str(x) binned <- x binned levels(binned$cats) <- substring(levels(binned$cats), 1, 4) binned > cats <- as.factor(c('2016-1','2016-2','2016-3')) > sales <- c(10,20,30) > x <- cbind.data.frame(cats,sales) > x     cats sales 1 2016-1    10 2 2016-2    20 3 2016-3    30 > str(x) 'data.frame':  3 obs. of  2 variables:  $ cats : Factor w/ 3 levels "2016-1","2016-2",..: ...