POTENTIAL CUSTOMER ANALYSIS USING K-MEANS WITH ELBOW METHOD

Abstract

This study aims to obtain cluster data of potential customers using the K-Means clustering approach supported by the elbow method to determine the correct number of clusters. The data sample that was processed was 100 customer data from a minimarket containing three criteria (gender, age, and purchase retention). The number of initial clusters is determined as 5 and then processed by calculating K-Means. The calculation of the SSE value in the K-Means process produces the lowest SSE value, and the sharpest elbow angle graph visualization is in cluster 4. So, it can be stated that the best number of clusters in this K-Means calculation is four (4) which are used as material for further analysis. Then the analysis results of four (4) clusters state that potential customers are those with high purchase retention, consisting of female customers who dominate in the three (3) clusters. Most potential female customers are customers with an age range above 35 years. Meanwhile, customers with less potential are spread across each cluster with varied gender and age but are not dominant. Thus, this knowledge can be used as a consideration for the management in determining the right promotion strategy.