Prediksi Korupsi Menggunakan Indikator Tata Kelola Publik (Public Governance) Berbasis Machine Learning

Corruption is a global problem faced by almost all countries and has a significant impact on development. Corruption is known to hamper economic growth by disrupting production efficiency and reducing innovation. Therefore, early detection of potential corruption is crucial.  The purpose of this study is to predict corruption levels using  public governance indicators through a machine learning approach. The models used include Random Forest Regression, Decision Tree Regression, XGBoost Regression, and Gradient Boosting Regression. The results show that  Random Forest Regression has the best predictive performance, as indicated by an R² value of 0.9802, higher than the other models. Furthermore, the results of the feature importance analysis indicate that the control of corruption dimension is the most dominant factor in determining the Corruption Perception Index (CPI) value. The predictive model allows for early warning of potential increases in corruption and helps focus improvements in governance by strengthening corruption control to support  sustainable development.

Keywords: Coruption predicition, random forest regression, decision tree regression, XGboost regression, gradient boosting regression