Multiple Testing P-Value Corrections in Statsmodels

[josef...@gmail.com]

Took me a bit longer than planned, but finally here is

http://jpktd.blogspot.ca/2013/04/multiple-testing-p-value-corrections-in.html


My multipletesting cleanup branch should go into master in a day or two.


The main thing that is still missing are bootstrap/permutation based methods.


Josef

[Vincent Arel]

very nice!

Vincent Arel-Bundock

[josef...@gmail.com]

Thanks, I hope it's helpful to get a rough overview over these methods.

Josef

[josef...@gmail.com]

statsmodels development can be slow

It was almost three years ago that I heard for the first time of these
multiple testing procedures from a request by a user of statsmodels.

Josef

[Skipper Seabold]

Know the feeling. We continue to make good progress though and things like this are super helpful (especially with the nice, extended tutorial-style documentation).

I think if you take number of lines of code (or number of what would be separate R packages) divided by developers we are much faster than other software ;)

Skipper

[josef...@gmail.com]

Yes, I think we are doing quite well fitting 21 years into a few calender years
http://www.ohloh.net/p/statsmodels/estimated_cost
(although I doubt these numbers apply to us, salary of 55,000 ?)

There are still some toughies out there waiting for the next years.

Josef

>
> Skipper
>

[Davide Cittaro]

Hi all,
I've previously ported p.value correction from R to python using this simple function:

https://bitbucket.org/dawe/utils/src/503fce8f6c4d6ca4a9c1a23a736a528d7fdf1fe5/fdr.py?at=default

I guess that would be obsolete now...

d

/*
Davide Cittaro, PhD

Coordinator of Bioinformatics Core
Center for Translational Genomics and Bioinformatics
Ospedale San Raffaele
Via Olgettina 58
20132 Milano
Italy

Office: +39 02 26439211
Mail: cittaro...@hsr.it
Skype: daweonline
*/