commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kim van der Linde <>
Subject Re: [MATH] Contributing
Date Wed, 11 Aug 2004 12:01:50 GMT

--- Eric MacAdie <> wrote:

> Phil Steitz wrote:
> >Kim,
> > 
> >Sorry for the response latency.  Thanks in advance
> for any contributions that you can make to [math].
> One thing to keep in mind is that we are trying to
> focus on commonly used stuff.  So, e.g. a basic GLM
> package would be more interesting than RMA.   Of
> course, if all is available....  It would be great
> to start by making the needed improvements /
> extensions to the current RealMatrix implementation
> to support a numerically sound multivariate
> regression / ANOVA package.  I would be less
> concerned about fitting the current architecture
> than laying the groundwork for good extensions.  We
> need to maintain backward compatability, but we can
> certainly extend as necessary.
> >  
> >
> What do "RMA" and "GLM" stand for?

RMA: Reduced Major Axis regression. 
Most people just use least squares regresion (model
1), but that assumes that the independent variable is
estimated with no or a relative low degree of error,
which is often not the case in biological sciences.
Model 2 regressions incorporate that variation as
well. There are three different approches for model 2
regressions, of which RMA is the most widely
applicable  as the two variables are allowed to be of
a different magnitude. Major Axis (MA) regression
rewuires that both vairables have the same magitude.
Generally, the methods try to minimize the variation
in both directions, not only in the Y-axis direction.

GLM: General Linear Model the superclass of all
regressions, (M)AN(C)OVA's and combination of those.



Do you Yahoo!?
New and Improved Yahoo! Mail - Send 10MB messages! 

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

View raw message