<![CDATA[Poverty in Indonesia remains a major development challenge that requires an understanding of socio-economic conditions across regions. Previous studies have applied machine learning for poverty classification. However, most have compared a limited number of algorithms and focused on district or municipality-level data. This study aims to compare the performance of four machine learning algorithms, namely Naïve Bayes, Support Vector Machine (SVM), Decision Tree, and Random Forest, in classifying the poverty levels of Indonesian provinces into “Low” and “High” categories. The study employs the CRISP-DM methodology and Synthetic Minority Over-sampling Technique (SMOTE) to address class imbalance. The dataset was obtained from Statistics Indonesia (BPS) for the 2018-2025 period and consists of five predictor variables and one label variable. The models were developed using RapidMiner with 10-Fold Cross Validation and evaluated using accuracy, precision, recall, F1-score, and balanced accuracy. The results show that Random Forest achieved the best performance, with an accuracy of 81.79%, an F1-score macro average of 80.46%, and a balanced accuracy of 81.11%. Furthermore, feature importance analysis of the Random Forest model was used to identify the relative importance of the indicators in distinguishing the poverty categories. The findings provide knowledge regarding the contribution of the indicators to the classification of provincial poverty levels in Indonesia.]]>
