EA SAVINGS ACCOUNT CUSTOMER CHURN PREDICTION AND CLUSTER PRIORITIZATION FOR XYZ BANK USING DATA MINING TECHNIQUE

• Main Author: Lyvtia Amandha, Keshya
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79291
• Institution: Institut Teknologi Bandung

PT Bank XYZ, Tbk (XYZ Bank) is one of the largest regional banks in Indonesia. XYZ Bank is focusing its efforts on developing their deposit products in order to increase third-party funds and transactional banking. One of XYZ Bank’s strategy is to retain their existing customer base using promotions and engagement campaigns. This strategy is considered ineffective, especially on their main deposit product, EA savings account. On June 2021, the churn rate for EA savings product is 13,02%, which is 83,5% of the total churns in XYZ Bank deposit products. One of the causes of this ineffectiveness is because XYZ Bank couldn’t identify potential churned customers. On the other hand, XYZ Bank have a large customer database that can be used to analyze their customer behavior. This analysis can be used to predict and prioritize churned customers that can be utilized to support XYZ Bank strategic actions to retain customers. This research uses data mining to build a prediction and cluster prioritization model for EA savings account churned customers. The Cross Industry Standard Process for Data Mining (CRISP-DM) is used as a framework to build the data mining model. The data mining algorithm used to build the prediction model are Random Forest and XGBoost, whereas K-means Clustering is used to build the cluster prioritization model along with the integration of AHP-RFM method. The data mining models obtained from this research are churn prediction model that is built with XGBoost algorithm, with 92,58% accuracy, 91,42% recall, and 65,42% precision, and cluster prioritization model, with a Silhouette score of 0,3953 and Davies-Bouldin index of 0,8208. Based on the evaluation, it is concluded that both models give good performances in predicting and determining cluster priority for EA savings account churned customers.