Sentiment Analysis of E-commerce Consumer Based on Product Delivery Time Using Machine Learning

Abstract

In this modern era, e-commerce sites, online selling, and purchasing are at the top of the list. Product quality and delivery time usually divert people’s sentiments about e-commerce. We conducted a sentiment analysis of consumer comments on Daraz and Evaly’s Facebook pages, and data were gathered from these two pages comments of Facebook. We evaluated the mood of client comments in which they expressed their opinions and experience regarding e-commerce pages services. With diverse models such as logistics regression, decision tree, random forest, multinomial naive Bayes, K-neighbors, and linear support vector machine in n-grams, we employ unigram, bigram, and trigram features. With 90.65 and 89.93% accuracy in unigram and trigram, random forest is the most accurate. With an accuracy of 88.49% in bigram, decision tree is the most accurate. Among the finest fits are the unigram feature and random forest.