[R] Function for gruping similar variables?

Timo W esuomikim at gmail.com
Thu Jul 8 22:38:59 CEST 2010


Does anyone have an idea on this?

On 6 Lip, 11:32, Timo W <esuomi... at gmail.com> wrote:
> Hi,
>
> I have a matrix of results of multiple 2x2 chi^2 tests, non-
> significant tests are marked as TRUE. Is there a function for grouping
> those variables in a similar way LSD.test from agricolae library does?
> I reviewed LSD.test's source but it's not helpful for me.
>
> This is my matrix:
>
>    [,1]  [,2]  [,3]  [,4]  [,5]  [,6]  [,7]
> 1  TRUE FALSE  TRUE  TRUE  TRUE FALSE  TRUE
> 2 FALSE  TRUE FALSE FALSE FALSE FALSE FALSE
> 3  TRUE FALSE FALSE FALSE  TRUE FALSE FALSE
> 4  TRUE FALSE FALSE  TRUE  TRUE  TRUE  TRUE
> 5  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE
> 6 FALSE FALSE FALSE  TRUE  TRUE  TRUE  TRUE
> 7  TRUE FALSE FALSE  TRUE  TRUE  TRUE  TRUE
>
> And I'd like to get similar output to this (LSD.test):
>
> Groups, Treatments and means
> a        oo      36.9
> a        ff      36.33333
>  b       cc      24.4
>   c      fc      12.86667
>
> For example, there's a group of variables 4, 5, 6, & 7 that don't
> significantly differ from each other. Is there any _simple_ way to
> mark all of them as a group?
>
> If you have any ideas, I'd be grateful.
>
> Thanks in advance,
> Timo
>
> ______________________________________________
> R-h... at r-project.org mailing listhttps://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guidehttp://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



More information about the R-help mailing list