commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Phil Steitz <>
Subject Re: [MATH] regression classes
Date Sun, 24 Oct 2004 03:58:36 GMT
Kim van der Linde wrote:
> Hi All,
> I have made three regression classes based on the Bivariateregression 
> class. The base class is an abstract LinearRegression class, with two 
> derived classes, LSRegression and RMARegression. Is there interest to 
> include these in the math package?
> I have changed some things, such as giving back a pair of doubles for 
> confidence intervals, adding significance testing, CI's for the intercept.
> Cheers,
> Kim


I think that both a more general regression framework and RMA regression 
in particular would be a good addition to [math].  What I would like to do 
is get 1.0 released with the "primitive" SimpleRegression implemented 
directly as it is.  Then we can talk about how to structure a more general 
framework for supporting different kinds of models. I am going to try to 
cut an RC2 this weekend, including the distributions changes and porting 
all of the RealMatrix changes to BigMatrix.

Are you OK holding off integrating this stuff until after 1.0?  I realize 
that the framework will likely require that we pull some of the 
implementation out of SimpleRegression, but we should be able to do that 
without breaking client code, so unless there is something big that I am 
missing, I would prefer to focus on getting the release out now.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message