<![CDATA[Gallstone disease is one of the most common gastrointestinal disorders, and early detection using non-invasive data can support screening prior to imaging examinations. This study aims to improve the performance of the Decision Tree algorithm in predicting gallstone disease by applying the Sequential Forward Selection (SFS) feature selection method. The dataset used is the public Gallstone dataset, consisting of 319 records and 38 features covering demographic, bioimpedance, and laboratory data. The research stages comprise Min-Max standardization, feature selection with SFS, and Decision Tree model development validated using 10-fold cross validation. Model performance was evaluated using a confusion matrix with accuracy and Area Under the Curve (AUC) metrics. The results show that the Decision Tree without feature selection (38 features) achieved an accuracy of only 67.71% and an AUC of 0.677. After applying SFS, the best performance was obtained using only 6 features, namely Coronary Artery Disease, Hyperlipidemia, Height, Hepatic Fat Accumulation, C-Reactive Protein, and Vitamin D, with an accuracy of 76.18% and an AUC of 0.762. Thus, SFS increased the accuracy by 8.46% and the AUC by 0.085 while reducing the number of features by 84%. These findings indicate that SFS not only improves the predictive performance of the Decision Tree but also produces a simpler and more interpretable model for gallstone prediction.]]>
Copyrights © 2026
