Customer Churn Prediction in Influencer Commerce: An Application of Decision Trees

Abstract

This study aims to predict customer churn in influencer commerce. As influencer commerce is a form of e-commerce, influencers directly sell products by uploading website links on their social media account after they promote products through SNS. The role of influencers is to promote brands and/or products on their social media accounts such as Twitter, Facebook, and Instagram. Recently, this role has expanded as a seller. This study implements the customer churn prediction based on the assumption that influencers have passionate support from their followers. The data collected by the influencer marketing agency in Korea from August 2018 to October 2020 includes the purchase details such as customer information, purchase item, and payment amount. In order to predict the churning customers, we apply the Decision Trees (DT) algorithm by using the computer software program, Rapidminer. Our analysis result shows the maximum prediction accuracy is 90% based on F-measure. This study contributes to customer churn prediction from the perspective of influencers. (C) 2021 The Authors. Published by Elsevier B.V.