Performance Evaluation of ML Algorithms for Diabetes Prediction

A serious condition that raises the blood glucose level is diabetes. If it is not treated, there are numerous consequences that could be prevented if it is detected sooner. The objective here is to study all the chosen algorithms and investigate the best one out of all. Here, we examine the modern machine learning classification strategies like decision tree, logistic regression, Naïve Bayes and support vector machine. The Pima Indians Diabetes Database (PIDD), that is made accessible through both Kaggle and the UCI repository, is the basis of experiments. Effectiveness of the algorithms are calculated using: accuracy, recall, precision, and F1-score. The accuracy calculated varies with each model and the one with a good accuracy is chosen as default for predicting the diabetes. According to the results, support vector machine outperforms other three algorithms with an accuracy rate of 83.