R子集数据帧的两个向量

Question

I have a data frame and two integer vectors named left and right .

I want to create a subset of the data frame in such a way that that the numbers in the vectors will indicate the range of the columns to be included in the subset.

For example, for the nth row in the data frame, I want to keep the values df[n,left[n]:right[n] . I tried doing so using mapply() :

aligned_rows<-apply(df,1,
                function(x) mapply(function(y,z)x[y:z], left, right))

But got an output that didn't make any sense.

Answer 1

This command does the trick:

as.data.frame(t(mapply(function(x,y,z) df[x,y:z],
                       x=seq_len(nrow(df)),y=left,z=right)))

Here is an example:

set.seed(10)
df <- data.frame(replicate(8,runif(4)))

#    X1    X2    X3    X4    X5    X6    X7    X8
#  0.51  0.09  0.62  0.11  0.05  0.86  0.41  0.77
#  0.31  0.23  0.43  0.60  0.26  0.62  0.71  0.36
#  0.43  0.28  0.65  0.36  0.40  0.78  0.84  0.54
#  0.69  0.27  0.57  0.43  0.84  0.36  0.24  0.09

Applying

left <- c(1,3,5,7)
right <- c(2,4,6,8)
as.data.frame(t(mapply(function(x,y,z) df[x,y:z],
                       x=seq_len(nrow(df)),y=left,z=right)))

yields

#    X1    X2
#  0.51  0.09
#  0.43  0.60
#  0.40  0.78
#  0.24  0.09

In order for that to work, each range defined by left and right must contain the same number of elements. Furthermore, both left and right must contain as many elements as there are rows in df .

Answer 2

As mentioned problem is not clear, hope below example will give some hints:

#dummy data
df <- data.frame(matrix(runif(20,1,50),nrow=4))

#right left dummy
right <- c(1,3,4)
left <- c(5,4,5)

#nth value, also try n <- c(2,4) to get 2nd and 4th rows
n <- 2

#return list of data.frames
lapply(1:length(right),
       function(x) df[n,right[x]:left[x]])

Answer 3

Without more information, your problem is ill-posed, because there's no guarantee that the number of items you want in each row will be the same. Remember that a data frame is a rectangular object, ie all the rows must have the same length.

What would be more reasonable to obtain is a list , which doesn't have this restriction:

mapply(function(l, r) df[, l:r], left, right, SIMPLIFY=FALSE)

Assuming this results in the same number of items per row, you can then combine them with rbind :

do.call(rbind, mapply(function(l, r) df[, l:r], left, right, SIMPLIFY=FALSE))

There are other issues, eg you're potentially combining items from different columns together which would be nonsensical if they have different classes. But you haven't mentioned this as a problem, so I'm going to assume that your data frame is really more akin to a matrix for which this kind of manipulation is more sensible.