In the digital era, e-commerce applications like Lazada facilitate online shopping for millions of users and gather reviews that assist both buyers and sellers. With the large volume of reviews, machine learning techniques, particularly sentiment classification, play a crucial role in automatically interpreting and classifying these sentiments. This study aims to apply and evaluate the effectiveness of the Support Vector Machine (SVM) model in classifying sentiments of Lazada app reviews. Utilizing text preprocessing methods, including the removal of stop words and vectorization using TF-IDF, the study successfully processed text data for the SVM model. Experiments were conducted with various test set sizes (10%, 20%, and 30%) to assess the model's performance under different conditions. The results show that SVM can classify review sentiments with the highest accuracy of 84.33% and an F1-score of 78.14% at a test size of 0.3. This research reveals that increasing the test set size contributes to better stability and accuracy in classification..