Implementation of Dynamic Mutual Information and Support Vector Machine for Customer Loyalty Classification

Abstract

Fast Moving Consumer Goods (FMCG) is known one of the important industrial sectors worldwide. It includes household and personal care products as well as processed foods and beverages. Because of the tight competition company must develop good marketing strategies. So, it is important for the company to know customer loyalty and also to predict the income as reference in company development planning. Data mining now is becoming popular technique for predicting customer loyalty. One of the well known data mining strategies is retaining customer’s strategy. In this paper, we would present a new model for predicting customer loyalty. The model is based on Dynamic Mutual Information and Support Vector Machine (DMI-SVM) to identify the relevant factors that affect the performance of the classification of customer loyalty. The comparison of two classification methods and several selected features is given to show the effectiveness of the methods. We validated the model by 10-fold cross validation method. Classification accuracy, precision, recall and f-measure are used to evaluate classifier performance on a test/hold-out sample. A result in this paper is shown that SVM method gives better performance accuracy than Naïve Bayes.