R Warning: Mistake in factor

Question

My R code is:

means_log_adj <- aggregate(lab_data[,delta_touse], 
    by = list(
        factor(mydata_adj$Response_EP, labels = c(\"non_responder\", \"re         


        

Answer 1

Adding to Andrie on Jun 20 '11 at 12:48

In the example provided as:

factor(x)
[1] a b c d e
Levels: a b c d e 

count the number of levels in the result set. Here there are 5 so you can set your ratio to 1:5. This works well when executing bubbleMap's key.entries variable. When I get this error, the first thing I do is what was prescribed by Andrie and update the ratio with the number of levels displayed.

Answer 2

After some trial and error I managed to reproduce your problem.

But let me start by saying that there is a very important difference between a warning and an error in R. When you report a problem, be sure to distinguish clearly between the two.

x <- letters[1:5]
factor(x, labels=LETTERS[1:10])

Error in factor(x, labels = LETTERS[1:10]) : 
  invalid labels; length 10 should be 1 or 5

This error occurs because you are telling factor() to re-label the data with levels that don't exist. I have specified 10 labels for a variable that only contains 5 levels. This means the labels and levels don't match.

There are two ways to fix this:

The first is to let R determine the levels, and simply call factor(x) without any parameters. (At a guess, this is probably what you should have done in your code.):

factor(x)
[1] a b c d e
Levels: a b c d e

The second is to call factor(x) and specifying the levels, not the labels:

factor(x, levels=letters[1:10])
[1] a b c d e
Levels: a b c d e f g h i j

---

You haven't provided sample data, so we can't test a solution. But try the following code:

means_log_adj <- aggregate(lab_data[,delta_touse], 
    by = list(
        factor(mydata_adj$Response_EP,), 
        factor(mydata_adj$sex), 
        factor(mydata_adj$timepoint)),
    mean)