An Excel© Macro for All Subsets Regression Using Genetic Algorithm Search

by Olcay Akman.

Abstract: Subset regression procedures have been shown to be more efficient than stepwise regression procedures. However when the number of predictors is large, these methods become cumbersome due to the high computational costs associated with the combinatorial nature of these algorithms. To alleviate this difficulty the use of Genetic Algorithms (GA) has been proposed an alternative to stepwise regression procedures. Genetic algorithms in an essence are computational search algorithm, commonly known as evolutionary computing, that utilize the mechanics of evolution by implementing the main principles of natural selection in searching for optima. Currently no main stream non-specialized statistical software provides an ability to run GA for all subset regression model selection. Here we created an Excel macro, which makes regression model selection via GA accessible to users of diverse fields. The tool is publicly available at www.ilstu.edu/~oakman.

Key Words: Stepwise regression, model selection, evolutionary computing

Author:
Olcay Akman, oakman@ilstu.edu

Editor: Graf, R.G.rgraf, rgraf@sunstroke.sdsu.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (112367 bytes.)
If you have any comments or for further discussion, contact an author.

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 1013 times since JULYE 2, 2013.


Return to the InterStat Home Page.