msoftrain - 10 months ago 39

R Question

Suppose a dataset where 10 participants have filled out a questionnaire with 3 items, x, y, and z.

`ss <- data.frame(x=rnorm(10), y=rnorm(10), z=rnorm(10))`

I want to compute the mean of these items for each person (personal mean).

I can easily do that this way:

`ss$personalmean <- rowMeans(ss[c("x", "y", "z")])`

or this way:

`ss$personalmean <- apply(ss[c("x", "y", "z")], 1, mean)`

But suppose I want to be able to do that in the with function. Then I run into problems.

This works:

`with(ss, (x+y+z)/3)`

But this doesn't:

`with(ss, rowMeans(c(x,y,z)))`

with(ss, apply(c(x,y,z), 1, mean))

I don't understand why. The example here is trivial, but in short, I would like to use the function

`apply()`

`with()`

Apologies if someone posted a good answer to that question somewhere, but googling what I'm looking for isn't easy given that "with" is a pretty generic term.

Answer Source

As @Frank mentioned in the comments, the `apply`

and `rowMeans`

work with matrix/data.frame. So, if we want to use `with`

, instead of `c`

, `cbind`

converts it to `matrix`

,

```
with(ss, rowMeans(cbind(x,y,z)))
with(ss, apply(cbind(x,y,z), 1, mean))
```