Data-Driven Analysis of Attendance Patterns in Group-Based Microfinance Programs: Evidence from Indonesia

Abstract

This study examines attendance patterns in Weekly Group Meetings (Pertemuan Kelompok Mingguan/PKM) within a group-based microfinance program. Attendance is not treated merely as a frequency indicator, but as a behavioral pattern influenced by both individual and group dynamics. Using a data-driven approach, this study applies K-Means clustering to identify attendance behavior patterns and Decision Tree classification to analyze socio-economic characteristics associated with these patterns. The results reveal three distinct participation clusters: active, moderate, and passive participation. The findings demonstrate that attendance behavior is heterogeneous and cannot be fully explained by measurable socio-economic variables alone. Instead, participation reflects a complex interaction between individual characteristics and group social dynamics. By integrating social participation theory with machine learning methods, this study provides a more comprehensive understanding of participation behavior in community-based microfinance programs.