Reputation: 4846
I have the following example data frame in R.
item index ptr
A 1 0.40
B 2 NA
C 3 0.30
D 4 0.35
E 5 0.44
F 6 NA
It is already sorted based on column = index
. Now, I would like to sort it by the column ptr
but by leaving the position of rows where ptr
= NA
intact. So the output I'm expecting is:
item index ptr
C 3 0.30
B 2 NA
D 4 0.35
A 1 0.40
E 5 0.44
F 6 NA
The regular df = df[order(ptr),]
doesn't work. Any ideas? Thanks much in advance.
Upvotes: 3
Views: 338
Reputation: 58835
ptr.na <- is.na(df$ptr)
order.idx <- seq(length=nrow(df))
ptr.idx <- which(!ptr.na)
order.idx[!ptr.na] <- ptr.idx[order(df[ptr.idx,"ptr"])]
df[order.idx,]
gives
> df[order.idx,]
item index ptr
3 C 3 0.30
2 B 2 NA
4 D 4 0.35
1 A 1 0.40
5 E 5 0.44
6 F 6 NA
Upvotes: 4
Reputation: 66852
try this:
> df
item index ptr
1 A 1 0.40
2 B 2 NA
3 C 3 0.30
4 D 4 0.35
5 E 5 0.44
6 F 6 NA
> df[!is.na(df$ptr), ] <- df[order(df$ptr, na.last = NA), ]
> df
item index ptr
1 C 3 0.30
2 B 2 NA
3 D 4 0.35
4 A 1 0.40
5 E 5 0.44
6 F 6 NA
Upvotes: 9