I am currently exploring two regression methods using kernels, namely Kernel Ridge Regression (KRR) and Support Vector Regression (SVR). I tune their parameters using a randomized grid search.
Using a polynomial kernel, KRR outperforms SVR, but when I use an rbf kernel it is the other way around.
What might be an intuitive explanation for this? Or should I interpret this as a specific trait of the dataset that I am analyzing?