Penumpukan Belanja di Akhir Tahun Berdasarkan Perspektif Karakteristik Belanja

Abstract

Government’s year-end spending spikes is a yearly phenomenon that could reduce the government expenditure quality. Therefore, it is necessary to anticipate through identifying the characteristics of expenditure which is in principle can be accelerated and prevent accumulation. This study aims to measure the variable components of government agency’s expenditure which can be used as an initial step in anticipating the year-end spending spikes using Random Forest Regression algorithm machine learning method with a Feature Importance approach. The results of the study shows that several expenditure variables that should not spike at the end of the year tends to accumulate at the end of the year while at the same time confirms procrastinating behavior. Through the development of existing models, the stakeholders can take advantage of these tools as an early warning on potential spike in spending at end of the year and map out recommendations for spending acceleration to realize quality spending.