2.3.1. SVM Flowchart

Figure 3 shows the flow diagram of the whole SVM model. First, the support vector machine has been implemented without any fine-tuning. Without fine-tuning, SVM takes regularization parameter C as 1, and, for the kernel, it uses the radial basis function (RBF). After that, the grid search has been applied to fine-tune the model. Then, different regularization parameters have been taken for the parameter combinations, such as values C, gamma values, and four types of kernels: the RBF, linear, poly, and sigmoid kernel. Also, 5-fold cross-validation has been applied to evaluate all possible combinations. Then, the model was trained again, and there was a significant improvement. The confusion matrix has been calculated based on this version.

Flowchart of SVM.