Implementation of the Spiral Optimization Algorithm in the Support Vector Machine (SVM) Classification Method (Case Study: Diabetes Prediction)

Abstract

Classification is a data mining method that is formulated to estimate group membership for data samples, this process is used to analyze the connections between data in a large data set. One of the classification methods that are often used is Support Vector Machine (SVM), in the SVM method there is a kernel function that helps in solving classification problems that cannot be separated linearly, one of which is the Radial Basis Function (RBF) Kernel. In using the SVM method with the RBF kernel function, Gamma and C parameters can affect the shape of the hyperplane in producing a good classification model, so that optimal Gamma and C parameter values are needed to produce a good classification. This study using the Spiral optimization Algorithm in optimizing Gamma and C parameters, by conducting several experimental stages in determining the best parameters of the Spiral optimization Algorithm to determine the Gamma and C parameters, SVM classification method with RBF kernel function can produce the highest accuracy is 86.15% with an average accuracy is 80.12% based on Pima Indians Diabetes dataset.