I’m learning programming/analysis with R. Here are a few things I’ve learned that were more painful than necessary.

• Suppose you have a formula such as fol <- formula(response ~ a + b + c). Every one of those values must be defined in your data frame – including the left-hand side (‘response’ in this case). This means you must add a column to your data frame for the left-hand side. You’ll probably do that with df$response<-rep(0,row_count_goes_here).
• If you call rpart with something like model <- rpart(fol, method="class", data=training), you might get Error in cbind(yval2, yprob, nodeprob) : number of rows of matrices must match (see arg 2). If all of your left-hand side values are the same, rpart throws this unhelpful message as its way of saying, “Hey! I can’t make a decision tree where all of the records (‘response’, in this case) have the same value! The decision is too easy.”
  • You’ll have to put the calculated value for the left-hand side into the table. Try something like training$response <- training$a + training$b + training$c
• to get the first 5 rows of a data frame with 13 columns use either:
  • df[c(1:5),1:13]
  • df[c(1:5,]