Optimizing Early Stage Diabetes Detection: A Robust Evaluation of Machine Learning Algorithms

The escalating global incidence of diabetes emphasizes the imperative for prompt detection to alleviate significant health adversities.This investigation assesses the efficacy and robustness of three machine learning algorithms-Decision Tree, Support Vector Machine (SVM), and Naive Bayes-utilizing methodologies such as Train-Test Split, K-Fold Cross Validation, and Stratified K-Fold Cross Validation.Critical performance indicators including Accuracy, Precision, Recall, F1-Score, and ROC-AUC were meticulously examined, with standard deviation employed to evaluate the stability of the models.SVM consistently surpassed the other algorithms, exhibiting superior accuracy and reliability across the various validation approaches, particularly within the context of Stratified K-Fold Cross Validation.Naive Bayes revealed commendable recall efficacy, while Decision Tree experienced augmented stability through the application of cross-validation techniques.The results underscore the significance of employing cross-validation methods, particularly Stratified K-Fold, for dependable model assessment in scenarios characterized by imbalanced datasets.Subsequent research endeavors should investigate ensemble methodologies and data augmentation strategies to further enhance the resilience of the models.