Financial Anomalies in Bank Subscriptions: Machine Learning and Bayesian Optimization for Term Deposit Analysis

The study uses the UCI Bank Marketing Dataset and applies the method of unsupervised anomaly detection to investigate customer behavior in the banking industry. In contrast to all other work done in the past, which includes dealing with predicting deposits, here we draw out the main factors of anomalous interactions, such as the job type, interaction duration, and contact timing. Methods like Isolation Forest, Local Outlier Factor, and One-Class SVM are applied to the imbalanced dataset to find anomaly patterns that show that managerial and self-employed jobs are related to specific anomaly patterns. Based on the results, it is suggested that demographic understanding must be integrated with the marketing strategies.