<![CDATA[Healthcare services constitute a crucial aspect in improving public well-being. Every individual has the right to receive healthcare services that are of high quality, safe, efficient, and affordable. This study aims to identify and analyze public perceptions and sentiments toward healthcare services at RSUD Soe, as well as to evaluate the performance of several machine learning methods in classifying such sentiments. The data were collected from 278 respondents through a Likert-scale questionnaire that represents perceptions and levels of satisfaction regarding various service aspects. Sentiment analysis was conducted using four machine learning algorithms, namely Naïve Bayes, C4.5, Random Forest, and Support Vector Machine. The results indicate that Naïve Bayes achieved the highest accuracy of 82.14 percent, followed by SVM at 80 percent, Random Forest at 79 percent, and C4.5 at 73.21 percent. This study also applied the Latent Dirichlet Allocation (LDA) method to identify the main themes within public feedback. LDA generated twelve topics reflecting key issues such as waiting time, availability of medical personnel, facility cleanliness, and the attitudes of healthcare staff. The majority of comments exhibited positive sentiment, particularly concerning staff friendliness and service quality. These findings were used to formulate improvement recommendations, including enhancing service quality, increasing the number of medical personnel, and optimizing facilities. This research demonstrates that a data-driven quantitative approach is effective in evaluating healthcare service quality and supporting more targeted decision-making. The results are expected to assist RSUD Soe in continuously and effectively improving service quality.]]>
