Data-driven Decision Making for Direct marketing of Banking Products with the use of Deep Learning and Random Forests

Nowadays due to the more customer-oriented era and increasing competition and advancing technologies, direct marketing becoming one of the most commonly-used marketing approaches worldwide. Many businesses, such as banks, apply direct marketing methods to reach more positive responses from their customers in order to minimize the campaigning cost and maximize the return on investment. To achieve this goal, banking and finance have to determine the target customer group to promote the bank product and services for them. It means that they need to know their customer attitude and characteristic individually to predict their needs and their reactions to product promotion. A huge amount of customer data has been stored in bank databases. This data as a rich resource can be analyzed to determine the most appropriate product to offer to each customer through the most effective channel. Since manually analyzing this data is not effortlessly feasible, this task should be executed automatically. In this work, a predictive model which is appropriate for bank product offerings was designed and built, which firstly classifies the customers to decide if they are interested in the product offering, and then clusters them for product and channel suggestions. This article combines data mining models with practical problems of the banking industry, and establishes a bank predictive response model and customer targeting, through Random Forest, Naive Bayes, and Neural network machine learning algorithms to classify customers and proposes corresponding suggestions for bank marketing based on the best classification in terms of accuracy and sensitivity of the result. The extracted classification rules and patterns can effectively help banks to divide customer groups with K-means clustering and take targeted measures to improve marketing efficiency.