Splitting triplicates into duplicates

问题

I am wanting to split sets of triplicate results into the three possible duplicates.

If the initial set is

    A   B   C
1 122 106 114
2 110 122 110

I want to transform it into

The combn function will do it line by line but I cannot figure out how to apply it across the full dataframe (which can be quite large, I only used two lines for demonstration purposes.).

回答1:

You can do the following:

Use lapply() to apply your combn for each row of the data
Use rbind to combine the results
Make use of the magic of do.call to combine the list created by steps 1 and 2.

In one line of code:

do.call(rbind, lapply(seq(nrow(dat)), function(i)t(combn(dat[i, ], 2))))
    [,1] [,2]
[1,] 122  106 
[2,] 122  114 
[3,] 106  114 
[4,] 110  122 
[5,] 110  110 
[6,] 122  110

回答2:

Here's another approach. I'll use data.tables to illustrate the concept. But it should be fairly straightforward using data.frames as well.

require(data.table) ## 1.9.2+
require(reshape2) ## for melt generic
setDT(dat) 
r1 = melt(dat, id="A", value.name="B")[, variable := NULL]
rbindlist(list(r1, dat[, list(B,C)]))
#      A   B
# 1: 122 106
# 2: 110 122
# 3: 122 114
# 4: 110 110
# 5: 106 114
# 6: 122 110

回答3:

Instead of iterating over all rows as suggested by Andrie, you can also iterate over all combinations returned by combn:

d <- read.table(text="A   B   C
1 122 106 114
2 110 122 110")

l <- apply(combn(3, 2), 2, function(x) setNames(d[x], c("A", "B")))
do.call(rbind, l)

##      A   B
## 1  122 106
## 2  110 122
## 11 122 114
## 21 110 110
## 12 106 114
## 22 122 110

This is probably faster than iterating over all rows. The function used in apply returns the columns of d currently of interest, and reassigns column names. Inspect l to see what's happening under the hood.

来源：https://stackoverflow.com/questions/25402561/splitting-triplicates-into-duplicates

标签

combinations