Open AccessClassification of Enchepalo Graph (EEG) Signals for Epilepsy Using Discrete Wavelet Transform and K-Nearest Neighbor Methods
Author(s) -
Maulana Angga Pribadi
Publication year - 2021
Publication title -
procedia of engineering and life science
Language(s) - English
Resource type - Journals
ISSN - 2807-2243
DOI - 10.21070/pels.v1i1.750
Subject(s) - pattern recognition (psychology) , electroencephalography , signal (programming language) , computer science , marketing buzz , epilepsy , wavelet , artificial intelligence , graph , k nearest neighbors algorithm , wavelet transform , psychology , neuroscience , theoretical computer science , world wide web , programming language
Epileps is a disorder of the contents of the nervous system of the human brain resulting in the presence of abnormal activity that is the excessive activity of neuron cells in the brain. In Indonesia there are more than 1,400,000 cases of Epilepsy each year with 70,000 additional cases each year. About 4050% occurs in children. A widely used method for assessing brain activity is through a sephalogram (EEG) Electrone signal. The Epilepsy classification system is built with extraction and identifikas stages. Wavelet exctraction is suitable for non-stationary signal analysis such as EEG signals. Wavelet tranformation can extract signal components only at the required frequency. So that it can also reduce the amount of data but without losing meaningful information. But to make it work and can be used on a system needs to be done classification in order to be able to distinguish between commands from each other. So it is used K-Nearest Neighbour (K-NN) classification method so that the signal that has been eliminated buzz can be directly entered into the classification to determine the correct wrongness of a data. In this study obtained the results of data accuracy value that K = 1 has the largest percent of 100% and the smallest percent is found in K = 7 and K = 11 namely 14.2% and 18.2% it is caused by the presence of classes that do not match the test data so as to reduce the percentage of accuracy in the K.