Sentiment Analysis on Application X on the Use of Red Oil Using the Naïve Bayes Method

Abstract

Red oil, as an alternative to traditional cooking oil, has gained public attention through reviews on App X. However, questions arise about how public sentiment is towards red oil and how the Naïve Bayes algorithm can classify positive and negative sentiments. This study aims to analyze user sentiment towards red oil using the Naïve Bayes method. The dataset used consists of 1,200 comments collected through the scrapping technique in 2024. After going through the process of removing duplicate comments, the number of data becomes 1,189. Before running the Naïve Bayes algorithm, the data is divided into test data and training data, with 238 data as test data and 951 data as training data. The analysis process involves pre-processing stages such as text cleaning, tokenization, and normalization, followed by word weighting with the TF-IDF method. The Naïve Bayes algorithm is applied for the classification of positive and negative sentiments. The results showed that 1,147 comments were positive sentiment, while 42 comments were negative sentiment with a total accuracy of 88.66%, then precision of 95.41%, recall of 92.44% and F1- 93.91% and it was found that the sentiment comments on the use of red oil had a greater positive polarity than negative polarity. This analysis provides important insights for producers and stakeholders regarding public perception of red oil, which is useful for strategic decision making, such as improving product quality and marketing campaigns. This method is expected to be a reference for further studies in the field of text classification and natural language processing.