Classification of Risk Attitudes from Customer Behavior with Machine Learning
Teeranai Sriparkdee, Prabhas Chongstitvatana · 2019
Every product and service in the market has its characteristic which has an impact on a consumer's decision to buy or use them. The risk is a distinctive characteristic of financial products, so in financial product and service design must use risk as a key factor. On the other hand, the consumer has different attitudes to the risk which can distinguish in 3 categories: risk aversion, risk neutral and risk seeking. Therefore, knowing risk attitudes of consumer who is the target market is an important key to define marketing strategy such as designing service and product, campaign, and promotion which going to be offered to them. Using the customer historical data, machine learning can be used to classify risk attitudes of each consumer. In this paper, we compare three machine learning methods to classify consumer's risk attitudes from their behaviors and identify important features. The results of the experiment show that the ensemble method, XGBoost, when used with resampling method ADASYN shows the best accuracy,