On regularization of generalized maximum entropy for linear models

Motivated by the advances in the estimation of parameters in linear models by regularization methods such as Ridge and Lasso regularizations, we investigate regularization of Generalized Maximum Entropy, which is an alternative estimation method in linear models. Our simulations confirm the better performance of the regularized Generalized Maximum Entropy estimation method, which could stimulate further theoretical research. An application of the new estimation method is illustrated with data from Thailand concerning the effect of education on economic growth.

Automated summary

Motivated by advancements in parameter estimation in linear models using regularization methods like Ridge and Lasso, the study focuses on regularization in Generalized Maximum Entropy, showing its superior performance and potential for further theoretical research. An application of the method is demonstrated with data from Thailand on the impact of education on economic growth.