Game theory in regression modeling: A brief review on Shapley Value regression

Article type: Research Article

Authors: Lipovetsky, Stan

Affiliations: Minneapolis, Minnesota, USA | E-mail: stan.lipovetsky@gmail.com

Correspondence: [*] Corresponding author: Minneapolis, Minnesota, USA. E-mail: stan.lipovetsky@gmail.com.

Abstract: The work describes developments in the multiple regression performed for building models resistant to multicollinearity, having meaningful robust solution for individual parameters, convenient for interpretation of the results, and good for prediction. A tool from the cooperative game theory, the Shapley Value analysis, have been tried for estimation of regression coefficients and relative usefulness of the predictors in a model. This approach has been checked and successfully applied in various real-life projects in data analysts for commercial companies. It is useful for decision makers in economics, management, marketing research, and any other practical fields.

Keywords: Regression modeling, resistance to multicollinearity, Shapley Value regression

DOI: 10.3233/MAS-210529

Journal: Model Assisted Statistics and Applications, vol. 16, no. 2, pp. 165-168, 2021