Use apply over rows

Question

I have a simple dataframe.

I would like to divide the entries in each row by the sum of all entries in that row, and get this:

  V1 V2 V3
1  0.08 0.33 0.58
2  0.13 0.33 0.53
3  0.16 0.33 0.50

It's pretty simple to use mydf[1,]/sum(mydf[1,]) and repeat it 3 times, but this is tedious.

I have an intuitive feeling that an apply function would work.

I have a vague notion I need to:
1) Put mydf[1,]/sum(mydf[1,]) in a functional wrapper
2) use apply(myfunction, 1, mydataframe)

But I'm not sure what the arguments to myfunction would be.

I'm a little confused on this and any help is appreciated.

Richard Herron · Accepted Answer · 2014-03-29T20:00:04.127

1

dat <- data.frame(matrix(1:9, nrow=3))
dat / rowSums(dat)

You could also use apply, but it will return a matrix (and require you to transpose with t).

t(apply(dat, 1, FUN=function(x) x / sum(x)))

edited Mar 29 '14 at 20:00

answered Mar 29 '14 at 19:53

Richard Herron

Sorry, my first solution was wrong, because it transposes. Fixed. – Richard Herron Mar 29 '14 at 19:56
So an `apply` function is unnecessary in this case. If I were to want to practice using `apply`, would this be a good exercise? – tumultous_rooster Mar 29 '14 at 19:58
You could use `apply` (I fixed my `apply` solution), but I think the `colSums` solution is little more transparent and saves a transpose and recast as data frame (`apply` returns a matrix). – Richard Herron Mar 29 '14 at 20:01
Great, thanks for your answer. I will study `apply` elsewhere. – tumultous_rooster Mar 29 '14 at 20:40
1

@MattO'Brien - I think it's worthwhile and makes code a little more readable and modular. Here's a great [summary](http://stackoverflow.com/questions/3505701/r-grouping-functions-sapply-vs-lapply-vs-apply-vs-tapply-vs-by-vs-aggrega/7141669#7141669). – Richard Herron Mar 29 '14 at 20:44
1

Interesting how that link starts off by subtly implying that the `pylr` package is probably better! – tumultous_rooster Mar 29 '14 at 20:48
`ddply` and `plyr` are fantastic. I tend to stick to base R when I can, but I don't have a good reason. – Richard Herron Mar 29 '14 at 20:50

1 Answers1