PRODUCT-PAIR RECOMMENDATION FOR CUSTOMERS USING MACHINE LEARNING TECHNIQUE

Abstract

Bank marketing campaigns mostly rely on human expert’s opinions on choosing potential customers. This method is time-consuming and lacks accuracy. Marketing offers being sent out to customers are mostly targeting the customers with irrelevant products that are not suitable according to their financials and transactional history. This mismatch means the bank is wasting their precious marketing resources and missing opportunities to please loyal customers and hence lost the profit. In this paper, a Pakistani banking dataset has been analyzed using two Machine Learning algorithms i.e. Decision Tree (Cart 4.5) and Random Forest to perform data analytics and recommend the best customer-product pair. The main intent of this research paper is to identify different banking products (Two Different Term Deposits Categories) suitable for various contrasting customers. The experimental results obtained will help in predicting and contacting the relevant customers with the best Term Deposit Category suitable according to their financials and also identifying the key features contributing towards the subscription of different Term Deposit categories. The performance evaluation of these algorithms are computed and determined by three of the statistical measures: classification accuracy, specificity, and sensitivity.