CODEWITHSUNDEEP
r
oercim
oercim

Reputation: 1848

R-combinations of columns of a data frame by partitioning columns

Let I have such a data frame(df):

col1  col2  col3
x1     x3    x4 
x2     y1    y2
y3     y4    y5

And let an another input data frame(dfi) is:

col1 col2
y    2
x    1

Desired output data frames are:

data frame1(df1):

col1 col2 col3
x1    y1   y2   
x2    y4   y5

data frame2(df2):

col1 col2 col3
y3    y1   x4
      y4

data frame3(df3):

col1 col2 col3
y3    x3   y2
           y5

Namely,

I want to create data frames where

How can I create such data frames using R? My original data frame is much larger and the second input data frame dfi can change. I know this is difficult problem. I will be very glad for any help. Thanks a lot.

Upvotes: 2

Views: 140

Answers (1)

agstudy
agstudy

Reputation: 121568

Not exactly what you are looking for( not the same column order). The problem is really very specific so it hard to explain the code but here some notes:

  1. Use col1 of dfi to find 'x' and 'y' columns in df. For x --> xi and for y --> yi. Put the result in a list like this one :

      List of 2
  $ :List of 3
     ..$ col1: chr "y3"
     ..$ col2: chr [1:2] "y1" "y4"
     ..$ col3: chr [1:2] "y2" "y5"
  $ :List of 3
     ..$ col1: chr [1:2] "x1" "x2"
     ..$ col2: chr "x3"
     ..$ col3: chr "x4"

  2. Use col2 of dfi , and list ll to find all combinations using Map and combin. We use also the helper function FUN to homogenize vector lengths.

Here the entire code:

ll <- lapply(dfi$col1,function(patt) lapply(df,function(x)x[grep(patt,x)]))
M <- max(dfi$col2)
## helper function to add '' if vectors don't have the same size
FUN = function(x){
  res <- lapply(x,function(y){
    if(length(y)<M)
      y <- c(y,rep('',M-length(y)))
    else y
  })
  as.data.frame(res)
}

ll_dat <- Map(function(x,m)combn(x,m,simplify=F,FUN),ll,dfi$col2)


Map(cbind,ll_dat[[1]],rev(ll_dat[[2]]))

# [1]]
# col1 col2 col3
# 1   y3   y1   x4
# 2        y4     
# 
# [[2]]
# col1 col3 col2
# 1   y3   y2   x3
# 2        y5     
# 
# [[3]]
# col2 col3 col1
# 1   y1   y2   x1
# 2   y4   y5   x2

Upvotes: 2

Related Questions