510 – Nonparametric Bayesian Predictive Models for Causal Inference
Agent-Based Model and Statistical Engineering of Automobile Purchase Behavior and Retail Actions
James Wendelberger
Urban Science
For several decades the automobile purchase behavior of new cars has been modeled and utilized by automobile manufactures for business decisions ranging from retail location, retail operations, consumer behavior, and competitive actions and reactions. Up until now these models have been optimized on an ad hoc basis. It is proposed that these models are actually interrelated and should be developed using a statistical engineering process. This process is demonstrated with an agent based proto type. The prototype displays unit sales and outlet profitability based upon outlet attractiveness and competitive actions and reactions on a daily basis. This prototype will serve as the base from which an all-encompassing state-of-the-art automotive market model is planned to be developed through a statistical engineering process. The entire model and real world data will adhere to the principals of the scientific method.