<![CDATA[People with Social Welfare Problems (PMKS) are a part of society that faces social, economic, and environmental challenges, thus needing proper support and action from local governments. However, identifying and assessing the vulnerability of PMKS at the Lubuklinggau City Social Service still encounters obstacles, such as the numerous social factors involved and the risk of bias in evaluations. Thus, this study intends to create a predictive model for PMKS vulnerability levels using a machine learning method based on Chi-Square feature selection and the Random Forest algorithm. The research starts with gathering and prepping PMKS data, which includes socioeconomic factors, family situations, and access to public services. The Chi-Square method is used to identify the most impactful features related to PMKS status. The findings show that access to public services, children’s education status, home ownership, and monthly income are the most important features, supported by the highest Chi-Square scores and very low p-values. These chosen features are then used as inputs for the Random Forest classification model. The experimental results reveal exceptional model performance, achieving accuracy, precision, recall, and F1-score values of 100% for both categories, specifically PMKS and Non-PMKS. These results suggest that combining Chi-Square feature selection and the Random Forest algorithm can yield a precise and reliable predictive model for classifying PMKS vulnerability levels. Therefore, the proposed model has significant potential as an objective and data-based support tool for the Lubuklinggau City Social Service in developing policies and ensuring better-targeted distribution of social welfare programs.]]>
Copyrights © 2026
