Generalized Maximum Entropy Estimation of Linear Models. (March 2017)

• Record Type: Journal Article
• Authors: Corral, Paul; Kuehn, Daniel; Jabir, Ermengarde
• Is Part Of: Stata journal. Volume 17:Number 1(2017)
• Page Start: 240
• Page End: 249
• Publication Date: 2017-03
• Subjects: st0473 -- gmentropylinear -- generalized maximum entropy -- maximum entropy -- linear; Statistics -- Periodicals; Statistics -- Computer programs -- Periodicals; 001.422
• DOI: 10.1177/1536867X1701700113
• Languages: English
• ISSNs: 1536-867X

Abstract

In this article, we describe the user-writtengmentropylinear command, which implements the generalized maximum entropy estimation method for linear models. This is an information-theoretic procedure preferable to its maximum likelihood counterparts in many applications; it avoids making distributional assumptions, works well when the sample is small or covariates are highly correlated, and is more efficient than its maximum likelihood equivalent. We give a brief introduction to the generalized maximum entropy procedure, present thegmentropylinear command, and give an example using the command.