Sentiment Classification of Customer Reviews in the Fast-Food Industry Using the Naïve Bayes Algorithm

In the digital era, online reviews have become a significant source of information, influencing consumer perceptions and purchasing decisions, particularly in the fast-food industry. This research focuses on classifying customer sentiment towards A&W restaurants based on online reviews using the Naïve Bayes algorithm. The objective of this study is to analyze customer feedback to understand their perceptions of A&W’s services and products. The research follows the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology, which involves six phases: business understanding, data understanding, data preparation, modeling, evaluation, and deployment. Data was collected from Google Reviews of the A&W Palem Semi branch, consisting of 200 customer reviews, which were preprocessed to remove irrelevant content and prepare the data for analysis. The Naïve Bayes algorithm was applied to classify the sentiments into three categories: positive, negative, and neutral. The model achieved an overall accuracy of 83%. However, the results revealed a significant class imbalance, with most reviews labeled as neutral. While the model performed well in identifying neutral sentiment (precision 0.89, recall 0.97, F1-score 0.93), it failed to classify positive and negative sentiments accurately, as both achieved precision, recall, and F1-scores of 0.00. This demonstrates that the data imbalance severely impacted the model’s ability to detect minority sentiment classes. The research concludes that while Naïve Bayes offers useful insights into customer sentiment, improvements are necessary, including applying data balancing techniques or exploring alternative algorithms such as SVM or Random Forest to enhance classification performance across all sentiment categories.