The effect of Data Augmentation Using SMOTE: Diabetes Prediction by Machine Learning Techniques

Diabetes mellitus, a severe and enduring condition characterized by impaired glucose metabolism, poses a substantial threat to public health. Its pervasive impact continues to escalate globally, with a rising incidence that challenges preventive measures. Despite earnest efforts, individuals struggle to evade the clutches of diabetes, necessitating innovative approaches for disease management. Traditional methodologies in diabetes health monitoring exhibit limitations, prompting the exploration of advanced techniques. This study employs machine learning (ML) methods to delve into diabetes, aiming to enhance diagnostic accuracy. The primary objective is to develop a method capable of precise diabetes diagnoses with a heightened level of precision. The investigation incorporates machine learning algorithms, specifically Random Forest (RF), K Nearest Neighbor (KNN), and Logistic Regression. The inclusion of these algorithms seeks to streamline data processing times.Notably, this study incorporates the Synthetic Minority Over-sampling Technique (SMOTE) as a data augmentation strategy. SMOTE addresses imbalances in the dataset, contributing to a more robust and representative sample. The research evaluates the effectiveness and accuracy of diabetes prediction using these algorithms both before and after SMOTE implementation. By considering the impact of SMOTE, the study aims to determine the optimal algorithm for assessing diabetes development. The comparative analysis sheds light on how SMOTE enhances the overall performance of machine learning models. This nuanced approach not only refines diabetes diagnostic protocols but also underscores the significance of addressing data imbalances in predictive modeling for enhanced precision in disease prediction.