Mobile Legends is currently one of the most popular MOBA type games with the highest number of players. This game occupies the top position in the top charts of the most popular games on the Play Store in the best-selling games category. There are many differences of opinion regarding the public's response or response to the mobile legends game as one of the most popular and best-selling games today. To sort and monitor these responses is not an easy thing because the number of responses or public responses posted on social media is very large if it is processed manually. Therefore, sentiment analysis is carried out in order to be able to monitor and sort responses or public responses quickly and automatically in categorizing responses that are positive, negative or neutral. The research data is taken from the opinion of the community of Twitter social media users on the mobile legends game and uses the Naïve Bayes Classifier (NBC) as a classification method with the addition of the Ngram feature extraction. This study resulted in 2,841 (12.8%) positive sentiments, 1,457 (6.6%) negative sentiments and 17,886 (80.6%) neutral sentiments. The results of the accuracy using the Naïve Bayes classification is 82% with the highest precision value of 85%, the highest recall value of 95% and the highest f-1 score value of 90%. Naïve Bayes feature extraction with Bigram has 78% accuracy, 87% highest precision, 89% highest recall and 88% highest f1-score. Extraction of the Naïve Bayes feature with Trigram resulted in an accuracy of 72%, the highest precision value of 89%, the highest recall value of 79% and the highest f1-score value of 84%.