Generalized maximum entropy estimation of linear models

• Paul Corral ^[3] ; Daniel Kuehn ^[1] ; Ermengarde Jabir ^[2]
  1. [1] Urban Institute

     Urban Institute

     Estados Unidos

     
  2. [2] American University

     American University

     Estados Unidos

     
  3. [3] The World Bank

  Mostrar afiliaciones +
• Localización: The Stata journal, ISSN 1536-867X, Vol. 17, Nº. 1, 2017, págs. 240-249
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • In this article, we describe the user-written gmentropylinear 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 the gmentropylinear command, and give an example using the command