Evaluation-Kernel-Using-Timeserie-Dataset

Support vector machine(SVM) is an advanced classifier for multidimensional data. To make SVM meaningful and effective for processing time series data, we need to select an appropriate kernel for SVM. An appropriate kernel for SVM relies on both a suitable kernel function and an appropriate distance metric. In this article, we will use dynamic time wrapping(DTW) as the distance metric to incorporate into different kernel functions(Cauchy, Gaussian, inverse multiquadric, Laplacian, Log and Rational quadratic) to compare the performance of different SVM kernel with DTW on time series classification.