Vasquez, Cesar; Dockstader, Patrick; Havertz, Brett; Phillips, Justin (Dixie State University)
Faculty Advisor: Chellamuthu, Vinodh (Dixie State University, Mathematics)
For any business, understanding the customer’s behavior is vital to maximizing income and minimizing costs. Our work aims to create an algorithm that analyzes the historical data from the customers and determines the target customers in an optimal way. We take on a data set from a transmission shop in California and seek out which factors produce higher potential for client value. We created a mathematical model that classifies the clients as low, medium, or high potential using this historical data. Furthermore, we demonstrated the model utility using the transmission shop’s data to compute the correlation of paying customers and customer history. The correlations are then used to create a conditional probability distribution which served to predict an expected rating score. Moreover, our results are validated by comparing the predicted ratings with the actual ratings in varying train and test cases from the data set. Our results show that the proposed algorithm is fast, simple, and intuitive, which could be utilized by the transmission shop in the future.