부산대학교 도서관

Machine learning and data mining help companies to build a tool that can make and take actions based on customer knowledge and information. Customer information is the base of maintaining long term relationship with customers and also known as customer relationship and management (CRM). Classification and segmentation of customer data set is utilised to maintain efficient relation with customers and subsequently increase the profitability and productivity. In this paper, author proposed customer segmentation based on demographic properties like gender, age and spending score and analysed the data set for interesting fact. The derived attribute data set is investigated for classification. Classification is used to categorise each customer into a number of classes, i.e., 'gold', 'silver', 'elite' and 'occasional'. Comparison of different classification algorithm is simulated by WEKA tool. Multi-layer perceptron (MLP) is found as the best classification algorithm with an accuracy of 98.33% compared to Naïve Bayes, regression and J48.