Technological developments create new ways for people to trade, especially the discovery of new technologies that make people want to have the technology to trade. Technological sophistication creates a new way of trading, namely with e-commerce applications, but there are conditions where the seller cannot know the level of satisfaction from his customers and also the problems experienced by his customers if only seen based on the rating in the case. in buying and selling smartphones. From these problems, a solution emerged to create a system that can filter out negative and positive comments, this study uses machine learning using the K-Nearest Neighbors, SVM, Naive Bayes algorithm with hyperparameters taken from previous research. The researcher uses the ensemble learning method with the Voting Classifier technique, which is an algorithm to combine several algorithms made. From the results of testing the highest accuracy was obtained by SVM with an accuracy value of 91.18% while the ensemble learning method got an accuracy value of 89.22% but for SVM the difference in the value of training and testing accuracy was 7.1% while for the ensemble learning method the difference in training accuracy and testing is 4%. The conclusion of the research objective is that the ensemble learning method can help improve the performance of the algorithm for commenting sentiment analysis on smartphone products.