Identification of Determinants of Inclusive Economic Growth Using the Metaheuristic Whale Optimization Algorithm Approach

Abstract

Inclusive economic growth demands the identification of key factors that drive equitable improvements in regional welfare. However, the complex interrelationships among social, economic, and demographic variables make traditional approaches insufficient for handling high-dimensional data. This study introduces an innovative approach by combining the Whale Optimization Algorithm (WOA) for feature selection with a Random Forest Regressor model to predict Gross Regional Domestic Product (GRDP) per capita as the main indicator of regional prosperity. The dataset consists of 210 regional observations and 18 independent variables. Feature selection using WOA was guided by minimizing the mean squared error (MSE), resulting in the identification of the 8 most relevant features. The retrained Random Forest model on the selected features achieved a high prediction performance, with an R² value of 0.9938 and a low RMSE. Furthermore, GRDP values were categorized into three regional welfare classes (Low, Medium, High), and the classification yielded 97.92% accuracy with high precision, recall, and F1-scores across all classes. These findings demonstrate that combining metaheuristic optimization and machine learning enables efficient and accurate identification of the key determinants of inclusive economic growth. The results provide valuable insights for formulating more targeted regional development policies.