R中的数据框列的组合

Question

I have a data frame that looks like the following:

df <-
Date  P1 P2 P3 P4
1/31  5  3  4  5 
2/28  2  3  7  8 
3/31  5  6  7  8 

I would like to take all combinations, by date, of columns P1...P4, taken 2 at a time -- 4C2. The result would be a list that contains the resultant data frames and will ideally look like this:

[[1]]
Date  P1 P2
1/31  5  3  
2/28  2  3  
3/31  5  6 

[[2]]
Date  P1 P3 
1/31  5  4   
2/28  2  7   
3/31  5  7  

.
.
.  

I'm trying to go about this using the combn() function. I can accomplish the combination by doing

combn(names(df),2,simplify=FALSE) 

but this gives me neither the dates nor the values in the result. How should I go about this task?

Answer 1

Try this

ind <- which(names(df)=="Date") # find the index of Date column
combn(ncol(df)-1, 2, function(a) cbind(df[ind], df[-ind][a]), simplify = F)

Answer 2

Using lapply you could do

split_vec = combn(names(df)[-1], 2)

lapply(1:ncol(split_vec), function(x) select(df, "Date", split_vec[, x]))