Performance Analysis of Naïve Bayes and Stochastic Gradient Descent-based SVM for Sentiment Analysis
Shrushti Jeetendra Vadher · International Journal of Computer Science and Engineering · 2025
This paper explores the effectiveness of sentiment classification on food review dataset, mainly focusing on Naïve bayes and Stochastic Gradient Descent-based Support Vector Machine (SGD-based SVM). The findings highlight the performance of both models and the impact of data preprocessing methods, as sentiment analysis is necessary for natural language processing and business customer services. The dataset obtained from Amazon on food review underwent preprocessing using Term Frequency-Inverse Document Frequency (TF-1DF) vectorization to transform textual data into numerical representations and Synthetic Minority Oversampling Technique to rectify class imbalances, ensuring fair and robust evaluation. The evaluation metrics demonstrate that SGD-based SVM has performed better than Naïve bayes with 84% and 79.9% accuracy, respectively. It can be observed that the SGD-based performs better as compared to the naïve bayes model.