Spatial Analysis of Random Forest Classification Model for Availability Mapping of Sports Facilities in Jakarta

This research analyzes the distribution of sports facilities in DKI Jakarta Province using spatial modeling and Machine Learning Random Forest algorithm in order to support Indonesia Emas 2045. The goal is to classify areas based on the level of availability of sports facilities into low, sufficient, and high categories, and evaluate the accuracy of the Random Forest algorithm in the classification. CRISP-DM methodology is used in this research, including Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment. The data analyzed includes spatial sub-district areas and attributes of sports facilities in DKI Jakarta. Random Forest was chosen because of its ability to classify complex data and identify feature importance. The results show that the distribution of sports facilities is uneven, with low categories more in Central Jakarta and North Jakarta, while high categories are scattered in other areas. Random Forest accuracy reached 89%, with high precision and recall in the high category.