Combination of XGBoost - Grid Search with SVM for Diabetes Diagnostics

Our study focuses on using data mining methods, such as XGBoost, Hyperparameter Grid Search, and SVM, to diagnose diabetes. Diabetes is a degenerative disease characterized by metabolic disorders and diabetic levels that exceed normal limits. In an effort to find the most effective model in diagnosing this disease, this study uses the available dataset and divides it into training data and test data with a ratio of 80%:20%. Through model testing, the accuracy results obtained for XGBoost without Grid Search were 87.0%, whereas with the implementation of Grid Search, the best accuracy obtained was 98.5%. In addition, the SVM model with the RBF kernel was also evaluated and provided an accuracy of 84.5%.