Machine Learning Approaches for Diabetes Prediction: Comparative Analysis and Pre-processing Insights

This study investigates the predictive capabilities of seven machine learning algorithms in the domain of diabetes prediction, revealing that the Support Vector Classifier (SVC) attained the highest accuracy of 0.793, closely followed by Logistic Regression with 0.787. Accuracy scores for Decision Tree, Random Forest, AdaBoost, and Gradient Boosting classifiers ranged from 0.727 to 0.78, while the K-Nearest Neighbors (KNN) model achieved an accuracy of 0.7. Notably, the SVM approach exhibited superior performance, particularly with a radial basis kernel function. Comprehensive dataset preparation, addressing issues such as missing values and varying class difficulties, was undertaken, with subsequent partitioning into training and testing sets via holdout validation. Various classification algorithms, including logistic regression, decision trees, K-nearest neighbor, and support vector machines, were employed. This research study highlights the importance of data preprocessing and the significant role of machine learning methods in improving the accuracy of diabetes prediction