Identification of Customers Satisfaction with Popular Online Shopping Apps in Saudi Arabia Using Sentiment Analysis and Topic modelling

Abstract

e-commerce platforms have evolved rapidly in recent years. They allow shoppers around the world to buy and sell goods and services over the Internet. Understanding the factors affecting the retention of existing customers in online shopping platforms and prompting their continued engagement is crucial to the success of such platforms. In this study, factors were examined through the lens of sentiment analysis, topic modelling and user reviews from three popular online shopping apps in Saudi Arabia, namely SHEIN, Noon, and Amazon. We employed sentiment analysis by implementing and comparing the performance of five well-known machine leaning methods on a large data set (55,285 user reviews). Stochastic Gradient Descent (SGD) was found to be the best performing classifier in terms of Macro F1 and accuracy at 91.88% and 92.71%, respectively. Afterwards, Latent Dirichlet Allocation (LDA), a topic modelling method was used to explore topics discussed by customers in the underlying dataset. Topics extracted from topic modeling application on the positive reviews were: fast and reliable delivery, easy shopping, quality of product /order/shopping, item price, good services, and item size. While the topics that emerged from the negative reviews were: poor services, refund/money issues, return products, product size, apps updates, and late delivery. The study adds to the understanding of e-commerce by using machine learning methods to identify various factors that influence consumers’ online shopping attitudes towards popular apps in Saudi Arabia. Moreover, such findings are valuable for e-commerce market to improve their services, increase customers satisfaction and the sales. We offered useful recommendations to e-commerce providers based on the results of this study.