Model Prediksi Risiko Kanker Serviks dengan Pendekatan Support Vector Machine

Cervical cancer is one of the leading causes of death in women, especially in developing countries due to delays in early diagnosis. Developing a risk prediction model based on the Support Vector Machine (SVM) algorithm is one way to support a more accurate and efficient early detection process. The research object is medical records of female patients obtained from hospitals in Medan City, with a total of 164 patient data. The development process was carried out through the CRISP-DM stages, which include data cleaning, feature transformation, class balancing with SMOTE, and dimensionality reduction using PCA. The evaluation results showed that the best model was obtained with a PCA configuration with 9 principal components (90% variance) and a test size of 80:20, resulting in an accuracy of 88%, a precision of 88%, a recall of 84%, and an F1-score of 86%. Cross-validation evaluation with 5 folds provided the best average performance and the smallest standard deviation, indicating model stability. The final model was implemented in a web-based system to facilitate digital early detection. This study shows that SVM with the SMOTE and PCA approaches is effective in predicting cervical cancer risk accurately and efficiently.