Comparison of Machine Learning Algorithms for Diabetes Classification

Diabetes is a chronic disease that influences how the human body turns food into energy and affects a significant portion of the global population. Diabetes prediction plays a major role in helping diagnose diabetes, which can help prevent complications and improve patient outcomes. This paper presents a machine learning approach for predicting the onset of diabetes using clinical and demographic data. The dataset used is from the Pima Indian Diabetes Database (PIDD), which was gathered from the National Institute of Diabetes and Digestive and Kidney Diseases. This dataset contains various variables such as pregnancies, glucose, blood pressure, skin thickness, insulin, body mass index (BMI), diabetes pedigree function, and age. Some preprocessing methods, such as oversampling and removing zero values to impossible data, was done to the dataset in order to balance and improve the input data. Then the paper compares four different machine learning algorithms, which are the K-Nearest Neighbors Algorithm (KNN), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), and Random Forest (RF). These models are then built to be predictive. The model’s performance is assessed using a number of metrics, including the F1-score, accuracy, precision, and recall. The Random Forest algorithm works better than other models, according to the results by achieving an accuracy of 89.5%, a precision of 97%, a recall of 83%, and an f1-score of 89%. This study shows the potential of machine learning approaches in diabetes prediction as well as the efficacy of the suggested preprocessing strategy which hopefully help clinicians identify individuals at risk of developing diabetes for early intervention.