Comparison of Machine Learning Models for Classifying Consumer Sentiment of Coffee Shops on Social Media X

With the intense competition in the coffee shop industry, understanding consumer opinions has become crucial for businesses. This study analyzes consumer sentiment toward the Janji Jiwa and Kopi Kenangan brands using tweet data from platform X. Sentiments were classified into positive, neutral, and negative categories using three algorithms: Logistic Regression (LR), Naïve Bayes (NB), and Support Vector Machine (SVM). The performance of these algorithms, in terms of accuracy and predictive capability, was evaluated using the TF-IDF method for text representation. The evaluation results show that LR achieved the highest accuracy at 79%, followed by SVM (78%) and NB (75%). Additionally, LR recorded consistent and balanced scores across the precision, recall, and F1-score metrics. These findings indicate that LR and SVM are more effective for multiclass sentiment classification in social media contexts