<![CDATA[User reviews play an important role in shaping perceptions of products, including the iPhone. Sentiment analysis of these reviews can provide valuable insights for companies to improve product and service quality. This study explores sentiment analysis of iPhone user reviews using a hybrid approach that combines RoBERTa and XGBoost to improve classification accuracy. The model was built and tested on a public dataset containing 2,960 reviews obtained from the Kaggle platform, following data cleaning processes. Preprocessing steps included handling missing values, encoding, and class balancing using the Synthetic Minority Over-sampling Technique (SMOTE). RoBERTa was used to extract text features and understand contextual meaning, while XGBoost served as the classification algorithm. The evaluation showed an accuracy of 99.74%, with an increase in the F1-score from 0.99 to 1.00 after applying SMOTE, particularly in the minority class. These findings demonstrate the superiority of the RoBERTa-XGBoost approach over traditional methods and contribute to the development of more balanced and adaptive classification models for imbalanced data.]]>
Copyrights © 2025
