User Satisfaction SentimentAnalysis For Mass Transportation Infrastructure (Case Study of Manggarai Station)

Evaluation of transportation infrastructure development is important considering that this is a solution from the government regarding the increasing mobility needs of urban communities. Social media and machine learning technologies enable an analytical approach to identify customer satisfaction in transportation infrastructure through sentiment analysis to accelerate policy evaluation analysis. The problem that arises is that it is necessary to provide specific data sources and appropriate analytical methods. To answer this, research was carried out which aimed to develop a sentiment analysis model and compare the accuracy of the Naïve Bayes, SVM, and KNN classifier algorithms in the sentiment analysis of Manggarai Station users as a case study. Data collection was carried out and sourced from Google Review. About 1453 review data were obtained which had gone through the preprocessing stage. The result of the analysis shows that the SVM classifier algorithm (RBF kernel) has the highest average accuracy with a value of 82%, followed by the KNN algorithm (k=12) with a value of 79%, and the Multinomial Naïve Bayes algorithm with a value of 75%. In the last three years, it was found that 64.28% of visitors to Manggarai Station had a negative impression of the station’s infrastructure facilities, another 27.53% had a positive impression, and 8.19% had a neutral impression.