Enhancing Predictive Modeling: Comparative Analysis of Decision Tree and Neural Network Approaches with Feature Selection

This research is focused on enhancing the efficacy of direct marketing for bank deposit products through data mining analysis. The primary challenge lies in accurately predicting and targeting prospective customers. The study conducts a comparative analysis between two machine learning techniques: Decision Trees and Neural Networks. The research findings reveal that the original Neural Network achieves an impressive accuracy of 96.20%, substantiating its proficiency in effectually classifying data instances. The precision score of 0.6344 contributes to precise and accurate positive predictions, and the F-Score of 0.5086 reflects a well-balanced performance across various evaluation metrics. The forward approach adopted in this study maintains a high accuracy rate of 96.23% and markedly improves precision to 0.6733 while upholding a harmonized F-Score of 0.4720. Both methods underscore the effectiveness of the Neural Network in meticulous data classification and the facilitation of accurate positive predictions. The original and forward approaches possess distinct merits, heightening predictive accuracy and fostering pattern recognition within direct marketing strategies for bank deposit products.