Predicción de deserción del cliente bancario mediante modelos de Machine Learning. Caso aplicado a un Banco de la región sur del Ecuador

Abstract

Abstract: The work developed shows the factors that promote churn within a financial institution. In this sense, it was identified that the Cross-Selling index, average semi-annual and quarterly deposits, and use of digital services and physical channels are the main predictive factors of churn. At the same time, it was found that customer engagement through the use of services and products offered by the bank acts as a loyalty barrier, increasing switching costs and reducing churn. The use of Machine Learning in the financial context allows for the promotion of evidence-based strategies to establish loyalty strategies and campaigns for customers at high risk of churn. The model with the best performance was XGBoost.