Comparative Studies of Intelligent Algorithms for Enhancing Machine Learning Training in Diabetes Prediction

Abstract

This The objective of this study is to compare the effectiveness of various intelligent algorithms in enhancing machine learning training for predicting diabetes from patient data. Early prediction of diabetes is crucial for preventing serious complications, and machine learning algorithms play an essential role in improving medical diagnostics. This research evaluates the performance of several algorithms, including Logistic Regression (LR), Random Forests (RF), Support Vector Classification (SVC), Gradient Boosting Machines (GBM), and K-Nearest Neighbors Classifier (KNN). These algorithms are compared based on multiple criteria: performance (precision, recall, F1-score, accuracy), computation time, model complexity, generalization capability, robustness, ease of implementation, and scalability. The study uses the Pima Indians Diabetes dataset, a well-known dataset containing several clinically relevant variables for diabetes prediction. The algorithms are evaluated using cross-validation methods, and regularization techniques are applied to optimize the hyperparameters.