Sentiment Analysis of Skincare Products Using the Naive Bayes Method

The number of reviews about skincare products can be used as an evaluation of product quality and satisfaction from consumers who have used it as well as considerations for other consumers to try the product. With the number of reviews, it is important to classify reviews into positive, negative, and neutral classes so that the level of product quality from each classification class can be known. The number of reviews causes the review classification process to be unable to be carried out automatically, so sentiment analysis is carried out. To determine the classification of positive sentiment, negative sentiment, and neutral sentiment on the skincare product, the Naive Bayes algorithm method is used. Naive Bayes was chosen because it is easy to implement and has a probability value to classify data. To determine the percentage of results from the specified classification, the Confusion Matrix will be used. The results of the classification process using the Naive Bayes method produce data into 3 types, namely 65 positive classes, 87 neutral classes, and 24 negative classes with an accuracy value of 73%, precision 77%, recall 61%, and f1-score 63%.