Analysis of Request for Quotation (RFQ) with Rejected Status Use K-Modes and Ward’s Clustering Methods. A Case Study of B2B E-Commerce Indotrading.Com

Abstract

In the current situation, PT. Innovation Success Sentosa with its B2B e-commerce platform, Indotrading.com, has a critical problem with the RFQ transaction process. Nearly 25% of all RFQs are rejected which results in the transaction being terminated, thereby losing potential profits. The purpose of this study is to find out, analyze and classify the characteristics of all rejected RFQs to find ways to overcome them. The analysis was carried out using the concept of data mining, using 3 algorithms, namely k-modes, average-linkage and ward’s clustering, which were applied to 7029 data objects in the rejected RFQ, during the period October to December 2021. To validate the result, we use the silhouette coefficient index. The clustering method that has the most optimal results, based on accuracy and consistency is the Hierarchical Ward's clustering method, with a value closest to 1, which is 0.664. Therefore, to solving the rejected RFQ, Ward’s algorithm is used as a reference. The clustering results show that the dominant RFQ was rejected, caused by sellers, with the reason ‘Stock not available now’ and dominant in the product categories ‘Electrical and Electronic supplies’ and ‘Construction and Property supplies’.