Application of the Naive Bayes and Decision Tree Algorithms Based on the RFM Model in Determining Customer Loyalty Segmentation (Case Study: Agung Denpasar Branch)

Abstract

Customer loyalty is an essential aspect in determining the sustainability of a business. Besides providing financial benefits, it can also help maintain a positive image of the product or service offered. This fact dramatically affects company performance, so we need a method that can help companies increase customer loyalty by segmenting the customer database as a whole. The purpose of the customer segmentation process is to determine customer behavior and create an appropriate marketing strategy to serve customers appropriately, and customers become loyal to the company. This paper aims to apply an accurate method of segmenting customer loyalty with several data processing models and algorithms. The Naive Bayes algorithm and the Decision Tree are used as classifiers in determining customer loyalty segmentation. The dataset is processed first using the ETL method (extract, transform, load) to produce a clean dataset without any error data attributes or features. The RFM (Recency, Frequency, Monetary) method was chosen as a suitable selection method for data attributes to get more optimal results in the classification process. The resulting dataset is classified into two classes or labels, namely Loyal and Non-Loyal. From the processes that have been tested, it can be concluded that the classification results have a relatively good level of accuracy. The dataset used is a database of customers who come to PT. Toyota Astra Motor, Agung Toyota Denpasar Branch from January 2019 to July 2020. The data will then be processed and classified into Loyal and Non-Loyal customer data, where the Loyal customer data is determined based on the customer who, in the past year, came twice and less than seven months of servicing the car. This research makes it easier for Data Analysts to segment or classify data in company databases because of its ease of implementation. The first study that combines data processing with the ETL, RFM method as well as the Naive Bayes and Decision Tree Algorithms for customer data segmentation.