Predicting the Propensity of Customers to Pay via Mobile Applications with Machine Learning Methods

The retail industry is experiencing constant transformation due to increasing customer needs and the rapid growth of innovative technologies. Getting to know your customers better comes to the fore in achieving these transformations. Improving shopping experience of customers by using personalized offers and purchasing options such as online wallets is usually one of the first applications in this journey. This study is about determining propensities of customers to use MoneyPay mobile application using their past spendings and behavioural patterns. Propensity modelling is a set of techniques for building predictive models that adopt past behavioural information to predict future actions of target audiences. In this study, we create attributes to represent tendencies of customers to use the MoneyPay mobile application. Using real-life data, we estimate the probability for each customer using various classification models such as Random Forest and XGBoost. We use various metrics to determine the success of the models and consider customer similarities for more reliable results. We present the results of our computational study.