Abstract:
Direct bank marketing campaigns often face low success rates, as only a small fraction of clients subscribe to term deposits. Traditional methods fail to capture the nonlinear patterns that influence customer behavior. This study used the Portuguese Bank Marketing dataset with 41,188 records (20 independent features and one target variable) to apply machine learning techniques for improving prediction accuracy and providing insights for more efficient campaigns. Preprocessing included handling missing values, detecting outliers, and balancing the target class with SMOTE, while exploratory analysis showed that middle-aged and educated clients, longer call durations, and favorable economic conditions increased subscription likelihood, whereas excessive contacts reduced effectiveness. Four supervised learning models-Random Forest, XGBoost, Decision Tree, and Logistic Regression-were evaluated using accuracy, precision, recall, F1-score, and Cohen's Kappa. Random Forest achieved the best performance with 91.42% accuracy and balanced recall and precision. XGBoost followed closely with 91.0% accuracy and the strongest precision (57.77%), though recall was weaker. Decision Tree achieved 88.75% accuracy with the highest recall (88.39%) but suffered from poor precision and overfitting. Logistic Regression performed the weakest, with 86.13% accuracy, serving as a baseline. The findings show that ensemble methods, particularly Random Forest and XGBoost, provide the most reliable predictions for term deposit subscription. By integrating preprocessing, exploratory analysis, model comparison, and feature evaluation, this research adds methodological, theoretical, and practical value. The results suggest that banks can improve targeting, reduce wasted efforts, and design more cost-effective marketing stsrategies using advanced machine learning models.
Page(s):
187-187
DOI:
DOI not available
Published:
Journal: 4th International Conference of Sciences “Revamped Scientific Outlook of 21st Century, 2025” , November 12,2025, Volume: 1, Issue: 1, Year: 2025
Keywords:
Machine learning
,
Bank Marketing
,
Classification Models
,
predictive analytics
,
Term Deposit