CUSTOMER CHURN PREDICTION IN THE BANKING INDUSTRY USING GENERALIZED LINEAR MODEL (GLM) AND ARTIFICIAL NEURAL NETWORKS (ANN)

• Main Author: Amanda, Della
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/74526
• Institution: Institut Teknologi Bandung

In a such competitive market, companies are making efforts to improve their business strategies. One of these strategies is to manage and strengthen interactions with its customers through Customer Relationship Management (CRM) so that the company's customers will continue using the products or services provided by the company and do not move to other competitors. This approach has the advantage of spending much lower cost than that needed in obtaining new customers, as well as being able to make the customers becoming long-term customers. One part of CRM is finding customers who have high probabilities of leaving the company or stopping using the company's products or services. This event is called customer churn. This final project focuses on predicting customer churn in a banking industry using two methodologies, namely: Generalized Linear Model (GLM) and Artificial Neural Networks (ANN). The comparison of the performances of the two methodologies is carried out using two measurement metrics, namely: the Area Under Curve (AUC) and the F1 score. The results of the performances comparison showed that the ANN provides a better ability to predict or classify customers who will leave the bank than those obtained using GLM. However, GLM could give an interpretation of the modeling results whereas the ANN method could not.