<![CDATA[Classification is one of the most researched topics by researchers from the field of machine learning and data mining. Machine learning methods that are often used include Support Vector Machine (SVM), Random Forest (RF) and Neural Network (NN). However, SVM does not always provide good accuracy. For example, when applied to highly imbalanced data, SVM will experience challenges. In addition, there is no single best method that can be applied to all classification problems. Currently, hybrid method approaches for data mining applications are becoming increasingly popular such as hybrid SVM-RF, SVM-NN and KMeans-SVM methods. In this study, a hybrid method of SVM-RF and SVM-NN was used to classify risk factors for anemia in pregnant women in Indonesia with a K-Means approach to cluster data misclassified by SVM. The results showed that the hybrid method can improve the performance of the SVM model. Hybrid SVM-RF provides a higher evaluation metric value compared to SVM-NN. The four evaluation metrics used, namely accuracy, balanced accuracy, sensitivity and specificity in SVM-RF are worth 0,989; 0,989; 0,988; and 0,989, respectively. The variables that contribute generally based on SHAP Global to the classification of risk factors for anemia in pregnant women in order are Age, Fe Tablet, Working Status, Education, Nutritional Status and ANC.]]>
Copyrights © 2025
