[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: good multivariate regression packages?



On 25 November 2005 at 13:43, Maciej Kalisiak wrote:
| Hello,
| 
| Can anyone recommend any good packages for doing multivariate
| regression?  I'm trying to model a function over a 5D domain (i.e., f:
| R^5 -> R).  If it has any bearing on the question, a cross-section of
| the function is shown in
|   http://www.dgp.toronto.edu/~mac/tmp/levelplot_disc=200_rad=0.3_k=5.png
| Other cross-sections look very similar, except that the shape is
| usually rotated.  The only other interesting thing about the data is
| that the samples never under-estimate the function (i.e., samples = fn
| + error, error >= 0 always).

You didn't say anything about the functional form of your model. 

If the model is linear in its terms [ and it would still be linear if it was
just a sum of nonlinear functions, polynomials, ... ] then Octave will do
fine. Or Python with NumPy / SciPy.

That said, you probably should look at R as it provides a real environment
for statistical computing, modeling, visualization, estimation, inference,
simulation, ...  I think there's also a full blown R / CRAN mirror at U of T
but I've forgotten where it is hosted.

| Ideally I'm looking for a package that would have Python bindings, but
| this is not necessary.  The main criterion is that it give decent
| results with relatively medium effort, for someone relatively
| unfamiliar with regression methods (i.e., no hardcore manual tweaking,
| etc.).

Sure, R can be driven from Python via RPy. And

	$ apt-get install python-rpy r-base

gets them both for you.

Greetings back to Ontario, and good luck,  Dirk

-- 
Statistics: The (futile) attempt to offer certainty about uncertainty.
         -- Roger Koenker, 'Dictionary of Received Ideas of Statistics'



Reply to: