Data Mining Approach to Predict Bank Saving Decisions Employing Classification Method

Abstract

This study aims to predict which individuals would opt for saving their money in a bank by looking at matters concerning their age, their current occupation, gender, and how much income they provided each month. The authors employ the Naïve Bayes classification method, which is a frequently observed yet occasionally straightforward enough approach to mine data. The very initial stages were a questionnaire via the web that got 109 responses. After a rigorous data clean-up procedure that included dropping features and grouping continuous variables, the dataset was split up. About 70% of it was used to train the model, and its remaining 30% was used to test to determine how well it performed in circumstances it hadn't experienced previously. The outcomes pointed out that the Naïve Bayes classifier exceeded the threshold of 81% accuracy, with precision and recall around 85%. These were attained because the model indicated a strong ability to identify the majority of positive cases (high recall), while also showing a tendency to predict positives slightly more often than necessary, leading to some false positives, as indicated by the confusion matrix. This suggests that the model prioritizes minimizing missed detections over avoiding errors. These findings contribute to banks that use this type of data mining classification method may be able to improve their marketing strategies, prioritize specific leads, make decisions more quickly while still obtaining results that are thought to be fairly accurate, or optimize services to make hesitant customers feel more likely to open accounts. But this relies substantially on how properly the model indicates the way individuals conduct themselves in their actual lives.