In today's competitive market, effective customer targeting through telemarketing campaigns is crucial for business success. This project explores the application of ensemble-based online machine learning techniques to enhance telemarketing success. The dataset comprises demographic and financial attributes of potential customers, including age, job type, marital status, education level, and loan status. The study compares an existing system leveraging Adaboost, Gradient Boosting, Random Forest, and XGBoost algorithms, though specific accuracy metrics are not disclosed, with a proposed system implementing Decision Tree, SMOTE for handling class imbalance, K-means clustering for segmentation, and a Stacking Classifier. The proposed system achieves promising results with Decision Tree achieving 81% accuracy, Stacking Classifier reaching 80%, and K-means clustering at 50%. Key findings underscore the effectiveness of ensemble learning in optimizing telemarketing campaigns, particularly in segmenting customer demographics and predicting subscription outcomes. The implications suggest that integrating these algorithms into online platforms can significantly improve campaign targeting and customer engagement. Future research directions may focus on refining ensemble techniques, exploring additional feature engineering methods, and integrating real-time data streams for dynamic campaign optimization.

KEYWORDS: Telemarketing, Ensemble Learning, Machine Learning, Predictive Modeling, Boosting Algorithms, Random Forest, XGBoost, Subscription Prediction, Marketing Optimization, Data Analysis.