Replicate each row of data.frame and specify the number of replications for each row?
I am programming in R and I got the following problem:
I have a data String jb, that is quite long. Heres a simple version of it:
jb: a b frequency jb.expanded: a b
5 3 2 5 3
5 7 1 5 3
9 1 40 5 7
12 4 5 9 1
12 5 13 9 1
... ...
I want to replicate the rows and the frequency of the replication is the column frequency. That means, the first row is replicated two times, the second row is replicated 1 time and so on. I already solved that problem with the code
jb.expanded <- jb[rep(row.names(jb), jb$freqency), 1:2]
Now here is the problem:
Whenever any number in the frequency corner is greater than 10, the number of replicated columns is wrong. For example:
Frequency: 43 --> 14 columns
40 --> 13 columns
13 --> 11 columns
14 --> 12 columns
Can you help me? I have no idea how to fix that, I also cannot find anything on the internet.
Thanks for your help!
Solution 1:
Update
Upon revisiting this question, I have a feeling that @Codoremifa was correct in their assumption that your "frequency" column might be a factor
.
Here's an example if that were the case. It won't match your actual data since I don't know what other levels are in your dataset.
mydf$F2 <- factor(as.character(mydf$frequency))
## expandRows(mydf, "F2")
mydf[rep(rownames(mydf), mydf$F2), ]
# a b frequency F2
# 1 5 3 2 2
# 1.1 5 3 2 2
# 1.2 5 3 2 2
# 2 5 7 1 1
# 3 9 1 40 40
# 3.1 9 1 40 40
# 3.2 9 1 40 40
# 3.3 9 1 40 40
# 4 12 4 5 5
# 4.1 12 4 5 5
# 4.2 12 4 5 5
# 4.3 12 4 5 5
# 4.4 12 4 5 5
# 5 12 5 13 13
# 5.1 12 5 13 13
Hmmm. That doesn't look like 61 rows to me. Why not? Because rep
uses the numeric values underlying the factor
, which is quite different in this case from the displayed value:
as.numeric(mydf$F2)
# [1] 3 1 4 5 2
To properly convert it, you would need:
as.numeric(as.character(mydf$F2))
# [1] 2 1 40 5 13
Original answer
A while ago I wrote a function that is a bit more of a generalization of @Simono101's answer. The function looks like this:
expandRows <- function(dataset, count, count.is.col = TRUE) {
if (!isTRUE(count.is.col)) {
if (length(count) == 1) {
dataset[rep(rownames(dataset), each = count), ]
} else {
if (length(count) != nrow(dataset)) {
stop("Expand vector does not match number of rows in data.frame")
}
dataset[rep(rownames(dataset), count), ]
}
} else {
dataset[rep(rownames(dataset), dataset[[count]]),
setdiff(names(dataset), names(dataset[count]))]
}
}
For your purposes, you could just use expandRows(mydf, "frequency")
head(expandRows(mydf, "frequency"))
# a b
# 1 5 3
# 1.1 5 3
# 2 5 7
# 3 9 1
# 3.1 9 1
# 3.2 9 1
Other options are to repeat each row the same number of times:
expandRows(mydf, 2, count.is.col=FALSE)
# a b frequency
# 1 5 3 2
# 1.1 5 3 2
# 2 5 7 1
# 2.1 5 7 1
# 3 9 1 40
# 3.1 9 1 40
# 4 12 4 5
# 4.1 12 4 5
# 5 12 5 13
# 5.1 12 5 13
Or to specify a vector of how many times to repeat each row.
expandRows(mydf, c(1, 2, 1, 0, 2), count.is.col=FALSE)
# a b frequency
# 1 5 3 2
# 2 5 7 1
# 2.1 5 7 1
# 3 9 1 40
# 5 12 5 13
# 5.1 12 5 13
Note the required count.is.col = FALSE
argument in those last two options.
Solution 2:
Nearly. You want to pass [
a vector of row indices, not row.names
. Try this...
jb[ rep( seq_len( nrow(jb) ) , times = jb$frequency ) , ]
rep( seq_len( nrow(jb) ) , times = jb$frequency )
# [1] 1 1 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
# [39] 3 3 3 3 3 4 4 4 4 4 5 5 5 5 5 5 5 5 5 5 5 5 5
Solution 3:
This might be more of a comment but seeing that all the other answers are suggesting new options - if you correct the spelling of jb$freqency
when creating jb.expanded, and convert jb$frequency
to an integer then the construction you mention in your question also works.
And why I suspect jb$frequency is a factor is because the incorrect frequencies are neatly ordered as 11,12,13,14.