Recommendation Product Based on Customer Categorization with K-Means Clustering Method

Abstract

Nowadays, web shopping is more than selling product. Many web shopping have basic analytics system to analyze customer data, transaction data including demographics, age and gender. They add many features in web shopping to maintain customer loyalty. In this research, we made a web shopping that can analyze customer shopping behavior. We used FM (Frequency and monetary) analysis based on a “transaction” data set. We categorize the customer based on how often they buy, how much they buy and how much the value of purchased item. We use K-Means algorithm to cluster the customer based on their transaction. Analyzing and understanding customers’ buying behavior can help the store to know what they are looking for. Therefore, at every customer web page that is in the same cluster will appear recommended products accordance with the transaction that has been done. Recommendation Products are presented is the prediction of the type of goods that may be chosen by the customer. So that the recommendation products that appear on web pages between customers will be different. The data used for this test is “Istana Accessories” store data from January to June 2014. The results show that recommendation product from K-Means algorithm successfully obtained and displayed on the customer page.