Forecast the Rating of Online Products from Customer Text Review based on Machine Learning Algorithms

Abstract

An online product's rating is an essential metric to understand the acceptability of that product to users. Shoppers use the rating to measure the quality and excellence of the online product. It helps an online shopper to decide to buy a product or not. It also helps the producer to further modification of that product during reproduction. Sometimes people buy a product online and give a text review of buying products but apathy towards giving a number rating, commonly a star rating. But producers need to know the rating of products for analysis of their business. Producers can use this rating for business analysis and can drive better revenue to their business. We have used some supervised machine learning approach to predict rating based on customer text review and compared the results between Random Forest Classifier, XGBoost and Logistic Regression algorithm with TF-IDF Vectorizer from extensive and series of experiments. We applied the algorithms mentioned above on the dataset named “GrammarandProductReviews” provided by Datafiniti. We analyzed the performance of each algorithm through the accuracy, precision, recall and f1-score. From the study, it is observed that Random Forest algorithm gained accuracy of 94% and precision, recall and f1-scores are 0.94, 0.94, and 0.94 respectively, which showed best compared to others.