Global search regression: A new automatic model-selection technique for cross-section, time-series, and panel-data regressions

In this article, we present gsreg, a new automatic model-selection technique for cross-section, time-series, and panel-data regressions. Like other exhaustive search algorithms (for example, vselect), gsreg avoids characteristic path-dependence traps of standard approaches as well as backward- and forwardlooking approaches (like PcGets or relevant transformation of the inputs network approach). However, gsreg is the first code that 1) guarantees optimality with out-of-sample selection criteria; 2) allows residual testing for each alternative; and 3) provides (depending on user specifications) a full-information dataset with outcome statistics for every alternative model.


Issue Date:
2015
Publication Type:
Journal Article
DOI and Other Identifiers:
Record Identifier:
https://ageconsearch.umn.edu/record/275931
ISSN:
1536-8634
Language:
English
Published in:
Stata Journal, 15, 2
Page range:
325-349

Record appears in:



 Record created 2018-08-16, last modified 2020-10-28

Fulltext:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)