A machine learning based approach for predicting upgrade of potential customers in banking sector

Abstract

Because of the outstanding growth in information, future predicting decisions about strategy development are extremely needed in each area. As a result, machine learning techniques have been used widely in bank- ing area. The main purpose of this thesis is that we employ a series of machine learning techniques to predict potential customers for future up-sell cam- paign, using knowledge development of databases in the banking indus- try. This thesis conducted in banking sector, it was aimed to reduce the cost for marketing to a minimum via analysis of potential customers for upgrading a product in the banks and estimate the expected pro t from model. Therefore, in order to select the best model, a comparison studies required to present the most appropriate approach to be used. This thesis presents a comparative method between three classi cation tech- niques of machine learning: Logistic Regression, Decision Tree and Ran- dom Forest. The evaluation is compared by Gain and Lift Chart and Kolmogorov-Smirnov chart after applying the two classi ers on a data set consisting of customer's personal, transaction and nancial informa- tion in Asia Commercial Bank in Viet Nam. Key words: Data Mining, Machine Learning, Bank, Customer Up- grade, Predictive Model Classi cation, Logistic Regression, Decision Tree, Random Forest, Python Language, Anaconda 3.