Data Mining Technique To Get Characteristics Customers of Bendesa Hotel With K-MEANS Algorithm

Abstract

This research aims to find customers based on characteristics of hotel customers who stay since there is still no research provides its technological state of the art. Through collaboration between Computer Science and Tourism, this research contributes on the development of K-Means Algorithm using WEKA application that can be elaborated into: 1) Search for best number of clusters used; 2) Identification of hotel customer characteristics; 3) Measurement of accuracy customer characteristics. This research can be used by hotel management to recognize customer characteristics so that they can develop strategies to get as many customers as possible, especially in Bali Province where Bali tourism is considered as one of the largest foreign exchange earners. K-Means algorithm uses CRISP-DM as a data mining life cycle which consists of 6 phases, the entire sequential phase is adaptive. The next phase in sequence depends on the output from the previous phase. In this research, it was tested on 2 clusters of up to 6 clusters. Using the value of sum of squared errors (SSE) is generated 5 clusters are the best from the other. Data on 5 clusters is used as reference to find characteristics of potential customers who stay in hotels. Through experiments, K-Means algorithm has an accuracy of 72% (108 of 150) tests using sample data compared to characteristics produced by K-Means. In the future, this research could be improved by: 1) collaboration between the K-Means algorithm and other clustering algorithms; and 2) add customer characteristics.