plyr in r: using wildcard grouping OR removing numbers and preserving characters in grouping column

Question

I am using plyr to calculate means and standard deviations in r. However, my grouping variable contains a combination of letters and numbers, so I need to either use some kind of wildcard in my grouping variable, or create a new grouping variable by removing the numbers from the original grouping variable. For example, with the following dataframe:

test5 <- structure(list(A = structure(1:6, .Label = c("JCT1", "JCT2", 
"JCT3", "LFR1", "LFR2", "LFR3"), class = "factor"), B = c(4L, 
5L, 3L, 7L, 3L, 6L), C = structure(c(1L, 1L, 1L, 2L, 2L, 2L), .Label = c("JCT", 
"LFR"), class = "factor")), .Names = c("A", "B", "C"), class = "data.frame", row.names = c(NA, 
-6L))

    A   B   C
1   JCT1    4   JCT
2   JCT2    5   JCT
3   JCT3    3   JCT
4   LFR1    7   LFR
5   LFR2    3   LFR
6   LFR3    6   LFR

I can use the following code to calculate means and sd:

library(plyr)
ddply(test5,~A,summarise,mean=mean(B),sd=sd(B))

which gives a result like

    A   mean    sd
1   JCT1    4   NA
2   JCT2    5   NA
3   JCT3    3   NA
4   LFR1    7   NA
5   LFR2    3   NA
6   LFR3    6   NA

However, I really need the groups to be JCT and LFR, so need to either 1) use a wildcard in the code (so groups are based on JCTand LFR, with the number being the wildcard), or 2) create a new column like C in my original dataframe that has removed the numbers from column A. So for example, if I could create this new column C then I could use the code

ddply(test5,~C,summarise,mean=mean(B),sd=sd(B))

to produce my desired result of

      C     mean          sd
1   JCT 4.000000    1.000000
2   LFR 5.333333    2.081666

Does anyone know of an easy way to do this? I thought I could use ifelse statements to somehow create a new column C, but this would require a lot of code as I have many different values in my real dataframe. I am hoping there is a quicker way.

Thanks!

Answer 1

Is something like this you are looking for?

library(plyr)
test5$A <- gsub('[0-9]+', '', test5$A)

ddply(test5, .(A), summarise, mean=mean(B, na.rm = T), sd = sd(B, na.rm = T))

    A     mean       sd
1 JCT 4.000000 1.000000
2 LFR 5.333333 2.081666

Answer 2

You could use regmatches and regexpr, to extract the letters and then summarize based on that

> ddply(test5,.(letter=regmatches(A,regexpr("[A-Za-z]*",A))),
    summarise,mean=mean(B),sd=sd(B))
  letter     mean       sd
1    JCT 4.000000 1.000000
2    LFR 5.333333 2.081666