R: Insert multiple rows (variable number) in data frame

Question

I have a data frame with, say, 5 rows, for 2 observables. I need to insert "dummy" or "zero" rows in the data frame so that number of rows per observable is the same (and can be bigger than N rows for longer one). E.g.:

#   This is what I have:
x = c("a","a","b","b","b")
y = c(2,4,5,2,6)
dft = data.frame(x,y)
print(dft)

  x y
1 a 2
2 a 4
3 b 5
4 b 2
5 b 6

Here's what I'd like to get, i.e. add N rows per observable to 4. Mock up df

x1 = c("a","a","a","a","b","b","b","b")
y1 = c(2,4,0,0,5,2,6,0)
dft1 = data.frame(x1,y1)
print(dft1)

  x1 y1
1  a  2
2  a  4
3  a  0
4  a  0
5  b  5
6  b  2
7  b  6
8  b  0

I started with getting the N rows in original data frame per observable with ddply, so that I know how many rows I need to add for each observable.

library(plyr)
nr = ddply(dft,.(x),summarise,val=length(x))
print(nr)

  x val
1 a   2
2 b   3 

# N extras will be 2 and 1 to reach 4 per obs. 

repl      = 4 - nr$val
repl_name = nr$x
repl_x    = rep(repl_name,repl)

print(repl_x)

[1] a a b
Levels: a b

dfa = matrix("-",nrow=sum(repl),ncol=1)
dff = data.frame(repl_x,as.data.frame(dfa))

names(dff) <- names(dft)
dft = rbind(dft,dff)
dft = dft[order(as.character(dft$x)),]

print(dft)

  x y
1 a 2
2 a 4
6 a -
7 a -
3 b 5
4 b 2
5 b 6
8 b -

I did achieve my goal, but in quite a few operations and transformations.

So, question - is there a simpler and faster way to insert arbitrary number of empty/dummy rows in several places in any data frame. Number of columns and rows can be any.

Note: the code above works, so I do believe this question is not a "review my code" type, but a genuine - "how to do it better" question. Thank you!

Answer 1

You can try using the "data.table" package which would let you use "length<-" to expand out your rows.

Demo:

library(data.table)
as.data.table(dft)[, lapply(.SD, `length<-`, 4), by = x]
##    x  y  z
## 1: a  2  2
## 2: a  4  3
## 3: a NA NA
## 4: a NA NA
## 5: b  5  4
## 6: b  2  5
## 7: b  6  6
## 8: b NA NA

---

Update

Upon provocation by Thela-the-taunter™, if you want to stick with base R, perhaps you can create a function like the following:

naRowsByGroup <- function(indf, group, rowsneeded) {
  do.call(rbind, lapply(split(indf, indf[[group]]), function(x) {
    x <- data.frame(lapply(x, `length<-`, rowsneeded))
    x[group] <- x[[group]][1]
    x
  }))
}

Usage would then be:

naRowsByGroup(dft, 1, 4)
#   x  y  z
# 1 a  2  2
# 2 a  4  3
# 3 a NA NA
# 4 a NA NA
# 5 b  5  4
# 6 b  2  5
# 7 b  6  6
# 8 b NA NA

---

Sample data:

x = c("a","a","b","b","b")
y = c(2,4,5,2,6)
z = c(2,3,4,5,6)
dft = data.frame(x,y,z)

Answer 2

dft = data.frame(x=c("a","a","b","b","b"),
                 y=c(2,4,5,2,6))

x <- 4 - table(dft$x)
dd <- rbind(dft, data.frame(x = rep(names(x), x), y = NA))
dd[order(dd$x), ]

#   x  y
# 1 a  2
# 2 a  4
# 6 a NA
# 7 a NA
# 3 b  5
# 4 b  2
# 5 b  6
# 8 b NA

And if you need, maybe you can add something if you have variables with >= 4 rows already like this

dft =data.frame(x=c("a","a","b","b","b",rep('c',6)), y=1)
x <- 4 - table(dft$x)
x[x < 0] <- 0
dd <- rbind(dft, data.frame(x = rep(names(x), x), y = NA))
dd[order(dd$x), ]

#    x  y
# 1  a  1
# 2  a  1
# 12 a NA
# 13 a NA
# 3  b  1
# 4  b  1
# 5  b  1
# 14 b NA
# 6  c  1
# 7  c  1
# 8  c  1
# 9  c  1
# 10 c  1
# 11 c  1

For an arbitrary number of columns:

dft = data.frame(x=c("a","a","b","b","b"),
                 y=c(2,4,5,2,6),
                 z=1,
                 zz=2)

x <- 4 - table(dft$x)
dd <- dft[1:sum(x), ]
dd[, names(dft)] <- NA
dd$x <- rep(names(x), x)
dd <- rbind(dft, dd)
dd[order(dd$x), ]

#   x  y  z zz
# 1 a  2  1  2
# 2 a  4  1  2
# 6 a NA NA NA
# 7 a NA NA NA
# 3 b  5  1  2
# 4 b  2  1  2
# 5 b  6  1  2
# 8 b NA NA NA