Diabetes Prediction based on Machine Learning Algorithms: MNB, Random Forest, SVM

The world's diabetes population is growing, which has concerned the medical community and prompted them to look for new ways to advance their technologies. In an effort to develop intelligent and successful diabetes prediction systems, research into machine learning approaches is still ongoing. The aim of this system is to help healthcare providers identify and effectively diagnose diabetes. The system created a machine model to forecast diabetes by utilizing a number of supervised machine learning approaches. This system explores the application of machine learning algorithms—Multinomial Naive Bayes (MNB), Random Forest, and Support Vector Machine (SVM)—in predicting diabetes using clinical and lifestyle data. The Pima Indians Diabetes Dataset from the National Institute of Diabetes and Digestive and Kidney Diseases was used to train and evaluate these models. In order to improve the model accuracy, data preprocessing, including feature scaling and handling missing values, was performed. After each algorithm was trained, its performance was assessed using measures including ROC-AUC, F1 score, recall, accuracy, and precision. Random Forest demonstrated high accuracy and robustness, efficiently handling complex interactions among features, while SVM showed strong performance in high-dimensional data settings. Multinomial Naive Bayes, although simpler, provided a baseline for comparison. The results suggest that Random Forest and SVM are effective tools for diabetes prediction, offering a promising approach for early diagnosis and intervention in clinical practice.