Application of Bayesian Additive Regression Trees to Analyze The Growth of United States Electric Automobile Industry

Annual Conference on Industrial and System Engineering 2019

工业技术（总论）

Ajidarma, P.^1 ; Irianto, D.^1

Industrial Engineering Study Program, Bandung Institute of Technology, Jl. Ganesha No.10, Coblong, Bandung

40132, Indonesia^1

关键词: Bayesian additive regression trees;    Economical factors;    Independent variables;    Marginal effects;    Market growth;    Multiple research;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/598/1/012035/pdf
DOI  :  10.1088/1757-899X/598/1/012035

学科分类：工业工程学

来源: IOP

【 摘 要 】

United States electric automobile industry has boomed significantly in the past recent years. Multiple research tried to analyze how a multitude of socio-economical factors correlate with and influence the growth of electric automobile market. This research aims to apply Bayesian Additive Regression Trees (BART) to study the relationship between each factor and the sales of electric vehicle, a proxy of the electric automobile market growth. The full posterior inference feature of BART enables the model to analyze the marginal effects of each predictors. The predictive BART models is selected based on the comparison of the in-sample fit and out-of-sample accuracy. Among sixteen independent variables, the model manages to identify key predictors and important interactions, which are consequential for the decision making in the electric automobile industry.