Siswandari Noertjahjani, Adhi Susanto, Risanuri Hidayat, Samekto Wibowo
Driven by a deep interest to find some spesific epilepsy EEG signal features as compared with normal ones. An array of electrodes, normaly the FP1, FP2, F7, F3, F2, F4, F8, C3, CZ, C4, T3, T4, T5, T6, P3, P4, PZ, O1, and OZ. The recorder signals were than processed and the standard sets of statistical quantities of means, variances, skewnesses, kurtosises, entropies, minima and maxima. Principal Component Analysis (PCA) were applied to these quantities to acquire two major one representing each quantity which separate best between epilepsy ictal and normal persons resorting to the SVM and KNN classification algorithms. The results show that the PCA elevates accuracy significantly and KNN achieves the mission better than SVM. © 2005 - 2015 JATIT & LLS. All rights reserved.
Electrical Engineering and Information Technology Dept., Gadjah Mada University, Yogyakarta, Indonesia; Electrical Engineering and Information Technology Dept., Gadjah Mada University, Yogyakarta, Indonesia; Neurology Dept., Gadjah Mada University, Yogyakarta, Indonesia; Departmentof Electrical Engineering and Information Technology, Muhammadiyah Semarang University, Semarang, Indonesia
