Machine Learning Techniques for Customer Satisfaction Prediction using Online Reviews

Abstract

Customer satisfaction is a key indicator of a successful modern business, making accurate customer satisfaction prediction an important task in Business Intelligence. With the rapid growth of online reviews, machine learning techniques can effectively analyze customer feedback and identify sentiment patterns associated with satisfaction. This study proposes an XGBoost-based machine learning (ML) approach for customer satisfaction prediction using the IMDb Movie Reviews Dataset containing 50,000 labeled reviews. The proposed methodology includes text preprocessing, spell correction, stop-word and punctuation removal, tokenization, TF-IDF feature extraction, z-score normalization, and an 80:20 training–testing split. The extracted textual features are subsequently provided to the XGBoost model for sentiment-based prediction. Experimental results demonstrate that the proposed model achieves 99.98% accuracy, 99.99% precision, 99.96% recall, and 99.97% F1-score, indicating highly effective classification performance. The results show that the suggested method can help businesses improve customer experience, happiness, and loyalty by providing trustworthy analysis of internet reviews.