PREDICTIVE ANALYTICS MODEL TO ENHANCE BANKING DECISION MAKING USING MACHINE LEARNING

Abstract

The present global economic crisis makes it difficult for banks to attract customers. Therefore, marketing is seen to be a useful technique for the banking industry to get clients interested in a term deposit. In banks, telemarketing is a commonly used kind of direct marketing. Customers seldom react favorably, therefore data prediction models may assist in identifying the most probable potential clients. Data mining helps direct marketing efforts succeed by foretelling which leads will sign up for term deposits. In this study, we used machine learning to the benchmark dataset of banking institutions' direct marketing campaigns to create an accurate classifier to forecast which consumer would accept a long-term deposit offer. Our research reveals the remarkable influence that machine learning methods may have on the outcome of a telemarketing campaign. Data preparation and model assessment are the two main phases. In the first phase, data must be cleaned by removing duplicate records and determining if missing values should be kept or removed, data visualization, and utilizing the response coding approach to encode category characteristics using label and one-hot encoding. The dataset is originally split into training and testing but the dataset is unbalanced so we needed to consider that while training so we used the balanced class weight approach and 10-fold cross-validation to solve the imbalanced class problem. The Random Forest algorithm is used for training and testing and a perfect classifier is achieved. The proposed system outperformed all the state-of-the-art techniques and achieved perfect classification.