Analisis Kesiapan Aparatur Sipil Negara (ASN) Menjelang Masa Pensiun di Provinsi Papua Barat dan Papua Barat Daya Menggunakan Machine Learning

Analysis of State Civil Apparatus (ASN) Readiness for Retirement in West Papua and Southwest Papua Provinces Using Machine Learning

Abstract

Every Indonesian civil servant (ASN) will eventually face retirement, yet readiness for this transition is rarely assessed individually, especially in West Papua and southwest Papua Province, where geographic dispersion and limited access to financial and skills-training services make uniform pre-retirement preparation inadequate. This study proposes a machine learning framework that combines predictive classification with clustering to support more targeted pre-retirement guidance. A Support Vector Machine (SVM) with an RBF kernel, applied to ten features selected through Mutual Information from questionnaire data covering financial, health, and psychological dimensions, was used to classify 1,500 ASN respondents into three readiness categories: Less Ready, Fairly Ready, and Ready. In parallel, K-Means clustering grouped respondents by post-retirement skill interests. The SVM model achieved 76.0% accuracy (macro F1-score 0.763, AUC 0.912) without indications of overfitting, while K-Means produced five clearly separated interest clusters (Silhouette score 0.4825). A chi-square test showed no significant association between skill-interest clusters and readiness categories (p = 0.9568), indicating that financial-psychological readiness and skills training should be addressed through separate, parallel programs. These findings offer agencies in West Papua a data-driven basis for designing segmented, targeted pre-retirement coaching programs.