[Biopython-dev] Documentation for Bio.LogisticRegression

[Jeffrey Chang]

Hi Michiel,

Wow, this is a really nice document!  About the only comment that I  
have, is that in the first sentence "distinguish K classes from each  
other" should be "distinguish 2 classes."  While there are multinomial  
models for logistic regression, this code does not handle them.

The "typecode" parameter allows the user to choose the type of Numeric  
matrix to use.  Since Newton-Raphson can eat up a lot of memory, on  
large problems, sometimes it may be beneficial to use single-precision  
floats rather than double, which is used by default.  Thus, "typecode"  
accepts Numeric typecodes, which are defined in the Numeric library  
like Numeric.Float16, Numeric.Float32, etc.  I left that parameter  
undocumented because 1) it goes deeper into the internals than I  
normally like to expose, and 2) I wasn't sure how useful it is.

Jeff



On Apr 25, 2004, at 8:00 AM, Michiel Jan Laurens de Hoon wrote:

> Dear all,
>
> Recently I have been using the logistic regression model in  
> Bio.LogisticRegression to predict transcription factors in bacteria  
> (thanks Jeff! Great work). Over the weekend, I wrote some  
> documentation for this module and submitted it to CVS. Jeff (or other  
> interested people), can you have a look at it and check if you agree  
> with my description? Feel free to add yourself as one of the authors.  
> By the way, the function train in Bio.LogisticRegression has a keyword  
> typecode, whose usage I didn't understand, so it is not included in  
> the documentation.
>
> The documentation is under biopython/Doc/cookbook/LogisticRegression:
>
> http://cvs.biopython.org/cgi-bin/viewcvs/viewcvs.cgi/biopython/Doc/ 
> cookbook/LogisticRegression/?cvsroot=biopython
>
> --Michiel.
>
> -- 
> Michiel de Hoon, Assistant Professor
> University of Tokyo, Institute of Medical Science
> Human Genome Center
> 4-6-1 Shirokane-dai, Minato-ku
> Tokyo 108-8639
> Japan
> http://bonsai.ims.u-tokyo.ac.jp/~mdehoon
>
> _______________________________________________
> Biopython-dev mailing list
> Biopython-dev at biopython.org
> http://biopython.org/mailman/listinfo/biopython-dev