Optimization of support vector machines by meta heuristic methods and applying on Parkinson’s disease data set

Abstract Parkinson’s disease occurs because of the decrease and insufficiency of dopamine cells over the time. Studies on computer-aided diagnosis systems that will help experts in making decisions for the early diagnosis of the disease remain up to date. Success has been achieved with computer aided diagnosis systems in a certain rate. The success rate varies according to the methods used, the data set and the optimization of the methods. In this study, parameter optimization of SVM classifier was performed to support the early diagnosis of Parkinson’s disease. The effect of different optimization methods that are current in the literature were compared on two different Parkinson’s disease data set. According to the results obtained, the highest accuracy rates vary according to the data set and optimization method. While Improved Chaotic Particle Swarm Optimization achieved high success in the first data set, Bat Algorithm achieved higher success in the other data set. While the success results obtained are better than some studies in the literature, they are at a level that can compete with some studies.