Performance Comparison of K-Means and DBScan Algorithms for Text Clustering Product Reviews

Abstract

The purpose of this study was to compare the accuracy performance of the K-Means and DBScan algorithms in clustering product reviews. This comparison evaluated to determine which algorithm is better in terms of accuracy. The two algorithms were chosen because they have different methods of clustering, K-Means uses centroid-based while DBScan uses density-based. Text clustering results can be implemented on e-commerce platforms, marketplaces or product review platforms. This can help customers in deciding what product they will buy. One of the factors that customers have difficulty in determining what product they will buy is the number of reviews that each product has, and the difficulty in concluding the advantages of each product that will be matched their needs or desires. With text clustering, it can be easier and faster for customer to determine whether the product is worth buying or not based on the product reviews they read. The data set used in this study is a review of the Cetaphil Facial Wash product from the Female Daily website. Firstly, data set goes through the Text Pre-Processing stage; then it will be clustered using two algorithms, K-Means and DBScan. After that, the results of the clustering of the two algorithms calculated for their accuracy performance and the performance results obtained. From the results of this study, it concluded that, in the review clustering of Cetaphil Facial Wash products, DBScan has 99.80% accuracy, which higher to compare with K-Means with only has 99.50% accuracy.