The Use of Machine Learning to Evaluate the Jatim Cerdas and Sehat Program: A Case Study in East Java

Abstract

This study evaluates the effectiveness of the Jatim Cerdas and Sehat Program using a quantitative approach based on machine learning to deliver quick, accurate, and data-driven insights. By combining sentiment analysis with Geographic Information Systems (GIS), the research uncovers public opinions, key issues affecting program implementation, and the spatial distribution of education and healthcare problems in East Java. The study employs a quantitative methodology, collecting data from Twitter through data crawling via the Twitter API. Sentiment analysis is conducted using the Naive Bayes algorithm, while spatial analysis utilizing GIS is employed to visualize the geographic spread of the issues. The sentiment analysis findings reveal that public perception of the program is largely negative. Key concerns identified include the unfairness of the education zoning system, confusing curriculum changes, high education costs despite promises of free education, and ongoing debates over the National Examination. In healthcare, complaints focus on disparities in healthcare facilities, inconsistent service delivery, and high treatment costs, including issues related to BPJS. Spatial analysis highlights priority areas, particularly rural and suburban regions, that require further attention regarding access to education and healthcare.