Comparison Of Kernel Support Vector Machine In Stroke Risk Classification (Case Study:IFLS data)

Abstract

Stroke s a disability main source and main disability source to lost years of disability-adjusted life. Currently the information technology development, especially the field of machine learning has an important role in early warning of various diseases, such as strokes. One of the methods used for stroke classifying is Support Vector Machine (SVM). In this study, we aim to compare several kernel functions in SVM such as linear, radial basis function(RBF), polynomial, and sigmoid for classifying stroke risk. We determine the best kernel based on accuracy, sensitivity, and specificity values. The result of this study shows that linear kernel function gives the best performance in classifying with values of classification accuracy 99.0%, specificity 100.0%, ,and sensitivity 97.0%. Those scores are the highest scores among the other kernel , that means the linear kernel function is the best method for classifying strokes risk.