# [R] manipulating "by" lists and "ave()" functions

Joshua Wiley jwiley.psych at gmail.com
Sat Jul 9 01:15:18 CEST 2011

```Hi Ivo,

See inline.

On Fri, Jul 8, 2011 at 3:42 PM, ivo welch <ivowelch at gmail.com> wrote:
> dear R wizards---more ignorance on my part, exacerbated by too few
> examples in the function documentations.
>
>> d <- data.frame( id=rep(1:3,3), x=rnorm(9), y=rnorm(9))
>
> Question 1: how do I work with the output of "by"?  for example,
>
>> b <- by( d, d\$id, function(x) coef(lm( y ~ x, data=x ) ))
>> b
>
> d\$id: 1
> (Intercept)           x
>     0.2303      0.3618
> ---------------------------------------------------------------------------------------------------
> d\$id: 2
> (Intercept)           x
>    0.05785    -0.40617
> ---------------------------------------------------------------------------------------------------
> d\$id: 3
> (Intercept)           x
>      0.269      -0.378
>
> getting the categories is easy:
>
>>  names(b)
>  [1] "1" "2" "3"
>
> but how do I transform the non-name info in this by() data structure
> into a matrix with dimensions 3 by 2?  (presumably, there is some
> vector operator that can do this kind of magic.)

Here is one option:

a <- do.call(cbind, b)

>
>
> Question 2:  Let's say I want to add only one of the two coefficients
> to d.  The naive approach
>  a <- ave( d, d\$id, FUN=function(x) coef(lm( y ~ x, data=x ))[2] )
> gives me the right coefficient in each row, but overwrites every
> entry.  I guess I can keep only the first column of a, and add it to
> d, but this seems a rather ugly and inefficient way.  How is this done
> better?

Do you actually want the coefficients duplicated for every row where
id is the same?

d\$coef <- a[, as.character(d\$id)]

>
> Question 3: repeat question 2, but keep both the intercept and the slope.

d <- cbind(d, t(a[, as.character(d\$id)]))

though you'll get a rowname warning

HTH,

Josh

>
>
>
> /iaw
> ----
> Ivo Welch (ivo.welch at gmail.com)
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> and provide commented, minimal, self-contained, reproducible code.
>

--
Joshua Wiley
Ph.D. Student, Health Psychology
University of California, Los Angeles
https://joshuawiley.com/

```