Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.



partition data, compute separate regressions, combine results
Posted:
Oct 5, 2012 5:00 PM


I have a dataset that is too large to run regressions using R (at least for the RAM I have).
Does anyone have advice on a way to get full regression coefficients from running regressions on subsets of the data?
Since each subset of data (assuming it is randomly selected from full data set) is an unbiased estimate of the full dataset, I guess I could just do repeated sampling and then average coefficients. I think this will get me a good answer but not an exact replica of the regression run on the full set of data set.



