<![CDATA[Thiis study aiims to compare the performanice of two deciision tree algoriithms, namely ID3 and Ci4.5, in classifying poverty status using household survey data from 2024. A quantitative experimental approach was employed, involving stages such as feature selection, data cleaning, transformation, model training, and performance evaluation. The ID3 algorithm was implemented manually based on eintropy iand information gaiin, while thei C4.5 algoriithm was implemented using thei scikit-learn library with parameter adjustments, including limiting tree depth and minimum samples per split to reduce the risk of overfitting. The idataset was divided iinto 70i% training and 30i% testing data. Performance evaluatiion used metriics including accuracy, precisiion, recalli, Fi1-score, and confusion matrix analysis. Results indicate that both algorithms can effectively classify poverty status, but the ID3 algorithm outperforms C4.5 in all evaluation metrics. However, C4.5 provides better model stability and a more simplified tree structure. Thesei fiindingis isuggest that the choiice of algoriithm should be taiilored to ithe dataset characteristics and analytical needs. This study contributes to the appliication of data miining techniiques in supporting poverty analysis and data-driven policy decisions, ensuring a more accurate and targeted distribution of social assistance.]]>
Copyrights © 2025
