Depending on what parts of their package you are interested in, to
implement a small subset of their functionality I recently wrote some code
in both Julia and R (to make sure I had an idea what they were doing to
replicate their results) based on the equations in page 258 of:
Hothorn T, Hornik K, Van De Wiel MA, Zeileis A, 2006. A lego system for
conditional inference. The American Statistician 60:257-263.
The base "independencetest" function I wrote only implements scenarios
where you have ytrans (one or more transforms of a dependent variable) and
one set of factors as a matrix of indicator variable (xtrans) although I
think incorporating more on the "x" side would not be hard. It does not
take into account multiple "blocks" and so far I have not needed to look
into how hard that would be.
On top of this I wrote functions to do the some of the transformations that
convert y to ytrans as input into the independenttest function (which is
surprisingly simple). The transformations I wrote make it possible to
implement the Kruskal-Wallis test for location, the Fligner-Killeen test
for scale, and a multisample Lepage test that incorporates those two tests
jointly (the reason I did this), all with the quadratic test (Cquad). It
would be trivial to write the transformations for many of the other
nonparametric tests. I wrote it to do both a parametric test as well one
using Monte Carlo. It appears to work well and reasonably fast with the
little testing I have done.
If you are interested I could send you the code.
On Thursday, February 11, 2016 at 5:27:26 PM UTC-7, Albert Cardona wrote:
>
> The formula operator is there via the GLM package [1], but I can't find
> the equivalent of the independent_test function [2].
>
> [1] https://github.com/JuliaStats/GLM.jl
> [2] http://www.inside-r.org/packages/cran/coin/docs/IndependenceTest
>
>
> Thanks in advance for any pointers.
>
> Albert
>
