Fuzzy-Silhouette Model for Automated Determining Loyalty Type User in E-Commerce

Abstract

The uniformity of marketing techniques in e-commerce causes failure in its application as a marketing communication strategy. This happens due to the lack of determination of the type of e-commerce user loyalty. Therefore, this study aims to automate the determination of user loyalty types in e-commerce using the Fuzzy-Silhouette model. In its implementation, this model has a range of types in each e-commerce 2-16 types with 4 numerical features that are processed, namely the length of transaction time, the distance between the last transaction and the data processing time, the number of transactions, and the total transaction. For each type of loyalty that has been generated, a standard deviation analysis was carried out to determine the characteristics of that type of loyalty. This research is a quantitative experiment conducted on three e-commerce sites, including UK Retail, Olist Store, and Alnafi Pakistan. The model evaluation results were obtained using the Davied-Bouldin Index (DBI) levels, respectively, of 0.31, 0.36, and 0.24, meaning that the smaller the DBI value obtained, the better the type formation results. So, it can be concluded that the validity of the Fuzzy-Silhouette model for automating the determination of user loyalty types in e-commerce has good results.