Feature analysis of customers purchasing cars in Japan

Abstract

Car sales in Japan are decreasing year by year. In addition, it will be affected by consumption tax increase and oil prices increase. Therefore, it is assumed that Car sales will be affected by external factor in Japan. Japan has problems with the declining birth rate and aging population. Car sales in Japan will be decrease caused by it, so it is necessary for Car industry in japan to know needs of customers. In this study, we analyze customer data from a car sales company “I” based in prefecture “A” in Japan to know customer needs. It is assumed that Cars in Japan have customers who purchase new or used cars, so their needs differ. Today, there are more customers buying new cars than customers buying used cars, however the opposite could happen for car industry affected by external factors in the future. So, it is important for car industry to know the needs of both customers. We use logistic regression analysis and decision tree analysis to know the difference between the two. After that, we evaluate the accuracy of AUC and confusion matrix as evaluation of both analyses. The analysis shows that customers who buy new cars are affected by the salesperson, and customers who buy used cars are less likely to be affected by the salesperson.