Support-Vector Machine applied in binary classification based on R. The reduced model can clarify a Pimas Indians Diabetes patient with 78.84% accuracy.
Support-Vector Machine is a supervised learning model applied in binary classification. SVM constructs a hyperplane to separate the binary data into two-part, which minimizes the misclassification error and maximizes the margin effectively. SVM can transpose the data into high-dimension space by kernels, then Fit a support-vector classifier in the enlarged space. With the diabetes dataset, we compared three kernel performances (radial, linear, and polynomial) and found that Linear-SVM is preferred with high accuracy and efficiency and is easy to interpret.