Predicting Diabetes Risk: A Comparative Analysis of Machine Learning Algorithms

Nowadays, diabetes mellitus is a worldwide health concern that is becoming more common. Therefore, for better early detection techniques (Machine Learning) is need of hour. Six machine learning algorithms are applied in this article to forecast the likelihood of diabetes in three stages: prediabetes, diabetes, and no diabetes. Three metrics i.e., recall, F1 score and precision are applied to evaluate the algorithms performance. Additionally, we have used the Synthetic Minority Oversampling Technique (SMOTE) for improving class imbalance. The results show that different algorithms and stages of diabetes function differently, which emphasizes the necessity of choosing algorithms according to the demands of certain applications.