Research on Bank Marketing Behavior Based on Machine Learning

Abstract

At present, under the background that data mining technology is becoming more mature and widely used in various fields, and due to the advent of the customer-oriented era and increased competition from banks, data mining technology is being widely used in the field of banking and finance to determine the target customer group And promote bank sales. Therefore, based on the Bank Marketing data in the UCI Machine Learning Repository database, this article uses the C5.0 algorithm to classify customers on the clementine experimental platform, and proposes corresponding suggestions for bank marketing based on the classification results. This article first explores and understands the Bank Marketing data set, and describes the distribution of the customer background in the data set. The quality of the data set was further explored, and the outliers and outliers were corrected by replacing them with normal data that were closest to the outliers or extreme values. This paper further selects the optimal feature variable. First, use the Filter node to filter the unimportant variables of the classification, and further select one of the more relevant variables to reduce the redundancy of the variables. The final variables are: previous, age, duration, outcome, contact, housing, job, loan, marital, education. Secondly, this paper uses sampling nodes to perform undersampling to balance the data set. On this basis, the C5.0 algorithm is used to establish a classification model and optimize parameters, and finally obtain eight classification rules. Based on this, suggestions are provided for target group determination. Finally, this article introduces the remaining four classification algorithms: C&T, QUEST, CHAID, Neural Networks, and compares the C5.0 algorithm with the four classification algorithms based on the balanced data set. It is concluded that several algorithms have certain differences and the overall prediction accuracy is good. This article combines data mining theory with practical problems of banking business, and establishes a bank target customer classification model based on C5.0 algorithm. The obtained classification rules can effectively help banks to divide customer groups and take targeted measures to improve marketing efficiency.