Performance Analysis of Supervised Machine Learning Algorithms for Diabetes and Breast Cancer Dataset

Machine learning has been very helpful in the early prediction of the fatal diseases such as cancer and diabetes for their timely treatment. The different supervised techniques such as linear regression, logistic regression, and decision tree, GaussianNB, SVC, Perceptron, MLPClassifier, Random Forest, Bagging Classifier, Gradient Boosting Classifier, Adaboost Classifier and Kneighbours Classifier have been used to predict breast cancer and diabetes. A comparison of various machine learning algorithms on the basis of parameters such as accuracy, precision, recall, MSE and f1-score has been done in this paper. It was seen that for the diabetes dataset the Logistic regression gave the best precision rate and best accuracy of 75.32% and the GaussianNB Classifier has the best recall, and the least MSE value. The best F1-score was obtained by GaussianNB and Gradient Boosting algorithms. For the breast cancer dataset SVC gave the best accuracy of 96.67% and highest precision, F1 score and the least MSE and MAE, the linear regression classifier has the best recall.