• Version
• Download 27
• File Size 370.89 KB
• File Count 1
• Create Date 12 March 2026
• Last Updated 12 March 2026

Predictive Modelling Framework Using Machine Learning for Bank Marketing Campaign Optimization

Prasad Kamble1                                                                       Prof.Kanifnath S Satav2

Student, MBA Department                                                           Professor, MBA Department

Dhole Patil College of Engineering, Pune                                       Dhole Patil College of Engineering, Pune

Abstract

Marketing campaigns are widely used by banks to promote financial products such as term deposits, loans, and investment services. However, traditional marketing approaches often involve contacting a large number of customers without accurately identifying potential subscribers, leading to low response rates and increased operational costs. With the rapid growth of data availability and advancements in machine learning, financial institutions can utilize predictive analytics to improve customer targeting and marketing efficiency.

This study develops a predictive modelling framework using machine learning techniques for bank marketing campaign optimization. The research utilizes the Bank Marketing dataset containing 41,188 customer records with demographic attributes, campaign interaction history, and macroeconomic indicators. The modelling process includes exploratory data analysis, domain-driven feature engineering, statistical feature selection, and a structured preprocessing pipeline using scaling and one-hot encoding.

Several machine learning algorithms including Logistic Regression, Random Forest, XGBoost, LightGBM, and K-Nearest Neighbors were evaluated using performance metrics such as F1-score, ROC-AUC, Average Precision, and cross-validation performance. To address class imbalance in the dataset, class-weight balancing techniques were incorporated into the modelling process.

Experimental results indicate that the Balanced LightGBM model achieved the best overall performance, providing the highest composite evaluation across F1-score, ROC-AUC, and Average Precision metrics. The proposed framework demonstrates how machine learning-based predictive analytics can significantly improve marketing efficiency, reduce campaign costs, and enhance customer targeting strategies for banking institutions.

Key Words: Machine Learning, Predictive Analytics, Bank Marketing, Business Analytics.