Perbandingan Metoda K-NN, Random Forest dan 1D CNN untuk Mengklasifikasi Data EEG Eye State

Abstract

The use of machine learning / machine learning methods is very important in developing identification of the status of the human eye, especially in terms of processing Electroencephalogram (EEG) signals to identify eye status. In previous research the method used can be a combination method between supervised learning and unsupervised learning, or a single method using supervised learning. In this study, the EEG Eye State classification uses a single method with supervised learning, namely using the following methods: K-nearest neghbors (k-NN), random forest, and 1D Convolutional Neural Networks (1D CNNs). The performance of the three classifier methods is measured using four measures, namely: accuracy, recall, precision, and F1-Score. From the experimental results it was found that the k-NN method has the best performance compared to the other two methods in terms of the four measures used, where the value of each measure is: accuracy = 82.30%; recall=82.30%; precision= 82.36%; and F1-Score=82.30%. K-NN is more suitable for classifying EEG Eye State than the other two methods, because all input attributes are from the dataset. has a data type of real numbers.