R help - Oct 2011 - Regression Package - for large dataset with lots of variables

Dear List,

I am trying to create a model for a relatively big dataset of a few million
obs. The number of variables is huge and runs into hundreds.
What are my choices for creating regression model - and what are the
drawbacks of using stepwise regression.

Is the BigLM package helpful, or should I try RevoScaleR or should I sample
and create model. What are other alternatives to stepwise regression for
computational efficiency.

I am on Ubuntu 64 bit Linux , and RAM is not a problem.

Regards,

Ajay

Websites-
http://decisionstats.com

	[[alternative HTML version deleted]]