A novel approach for the hyperparameters of support vector regression

In order to determine the hyperparameters of support vector regression (SVR), an approach with a two structured method is proposed to determine the kernel parameter σ and ε in the ε-insensitive loss function. Firstly, the kernel parameter σ of a Gaussian kernel function is determined by the competitive agglomeration (CA) clustering algorithm. The CA clustering algorithm incorporates the advantage of both hierarchical and partitioned clustering algorithms. Besides, it can find the nearly "optimum" number of clusters as well as its center of clusters in the clustering process. Secondly, the repeated SVR approach is proposed to obtain a proper ε in the ε-insensitive loss function that can be included in most of the data. Based on the efficiently structured way for choosing the hyperparameters σ and ε, the simulation results have shown that the proposed approach comes close to the "optimum" hyperparameter region