Clustering Underprivileged Student Aid Recipients Using Fuzzy C-Means

Abstract

Social assistance distribution in boarding-school educational settings requires a consistent mapping of prospective recipients so that the verification process does not rely solely on manual assessment. This study aims to cluster the socioeconomic profiles of students as a supporting basis for determining aid verification priorities using Fuzzy C-Means and to present the results through a Streamlit application. The research data consisted of 843 records of female secondary-level students from the 2022 cohort, obtained from the boarding school data management unit. After data cleaning by removing inactive student records, 781 records were processed using the attributes of number of siblings, parents’ occupation, and parents’ income. Categorical data were transformed into numerical codes and clustered using two clusters, a fuzziness parameter of 2, a maximum of 100 iterations, and a tolerance value of 0.00001. The Fuzzy C-Means process converged at the 28th iteration and produced Cluster 1 with 635 students and Cluster 2 with 146 students. The Silhouette Coefficient value of 0.4800 indicates a moderate level of cluster separation. The Streamlit application displays the initial data, preprocessing results, data transformation, and clustering results interactively. The findings indicate that Fuzzy C-Means can be used as a supporting tool for mapping prospective student aid recipients. However, clustering results should not be used as an automatic decision; aid recommendations must still involve document verification and staff assessment to prevent mistargeting