In order to perform the classification, we used three classifiers: KNN, RF, and SVM. Three different classification approaches were tested to evaluate the classification model. Three classifiers were compared in terms of the classification performance parameters to select a suitable classifier for a provided EEG data set. The KNN classifier, also called the lazy learner algorithm, assumes similarity between available data and new data and assigns the most similar class [55]. The KNN algorithm calculates the distance by utilizing the distance measure, e.g., Euclidean and Manhattan distance measures. In this work, the Euclidean distance measure was used for K = 1, 3, and 5. The SVM is a commonly used linear classifier that utilizes the hyperplane technique to maximize the distance from the nearest training datapoint to easily identify classes [56]. The SVM can predict good accuracy based on one of its abilities to select a suitable kernel function. We used the Pearson VII function-based universal kernel function. The RF is an ensemble learning technique that utilizes the concept of bagging, constructs a collection of decision trees, and takes the
