A Comparative Study to Predict the Diabetes Risk Using Different Kernels of Support Vector Machine

Diabetes is a metabolic issue wherein there is an increment in the level of glucose in the blood. The principle explanation behind diabetes is less emission of insulin from the pancreases. After some time, high blood glucose levels can harm the body's organs. This research study aims to predict diabetes for a patient with a higher exactness by consolidating the aftereffects via Machine Learning methods including Support Vector Machine (SVM). In this system, four types of kernel Linear, RBF, Polynomial, Sigmoid are used. From Sigmoid, the best Kappa Statistics (98%) is gained whereas polynomial kernel has shown the lowest performance. From the perspective of Accuracy, Precision, Recall, F1-Score, Specificity and ROC area Sigmoid kernel is also given the best outcome rather than other kernels. So, in this research study, out of these four kernels, the Sigmoid is the most preferable for predicting diabetes disease.