R , how to stack and keep factor

Question

R question: How do I stack two or more columns of numbers and keep a factor as well

I have a data.frame like this:

patient analyte1value analyte2value analyte3value  
pt1 1 3 5  
pt2 2 6 7  
pt3 9 10 2  
...  

I know I can use stack(select=c(columnnames)), but I lose the patient factor.

I want to get out:

pt1 1 analyte1  
pt1 3 analyte2  
pt1 5 analyte3  
pt2 2 analyte1  
pt2 6 analyte2  
...

(I have a sneaking suspicion that I need plyr or something like that...)

thanks.

Answer 1

One option is one of Hadley's other packages: reshape2:

> require(reshape2)
> dat
  patient analyte1 analyte2 analyte3
1     pt1        1        3        5
2     pt2        2        6        7
3     pt3        9       10        2
> melt(dat, id = "patient")
  patient variable value
1     pt1 analyte1     1
2     pt2 analyte1     2
3     pt3 analyte1     9
4     pt1 analyte2     3
5     pt2 analyte2     6
6     pt3 analyte2    10
7     pt1 analyte3     5
8     pt2 analyte3     7
9     pt3 analyte3     2
> str(melt(dat, id = "patient"))
'data.frame':   9 obs. of  3 variables:
 $ patient : Factor w/ 3 levels "pt1","pt2","pt3": 1 2 3 1 2 3 1 2 3
 $ variable: Factor w/ 3 levels "analyte1","analyte2",..: 1 1 1 2 2 2 3 3 3
 $ value   : int  1 2 9 3 6 10 5 7 2

One can do this in a more long-winded fashion using reshape() from base R:

reshape(dat, direction = "long", sep = "", varying = 2:4,
        times = names(dat)[2:4], idvar = "patient",
        timevar = "variable", v.names = "value")

with the main difference being that variable isn't a factor with base reshape(). I presume the user-unfriendliness of that was a motivation for writing reshape2...

Answer 2

If I understand correctly, you want to reshape your dataframe to a long format.

reshape(df,varying=list(2:4),times=names(df)[2:4],
  idvar="patient",v.names="value",direction="long")