Although the cognitive parts of their brains are intact, some individual scan only interact with the outside environment through eye movements. Those people suffer from severe motor disabilities preventing them from moving all their limbs. Recently, Human Computer Interfaces (HCI) has emerged to help these people by providing them a new way for communication. These interfaces are based on detecting eye movements. Electro-oculogram (EOG) records eye movements through few electrodes placed around the eyes vertically and horizontally. In this paper, EOG vertical and horizontal signals are analyzed to detect four eye movements (left, right, up and down) along with blinking. Three statistical features are extracted from filtered EOG signals. Extracted features from horizontal and vertical EOG signals are concatenated to form final feature vector. K Nearest Neighbor (KNN), Linear Discriminant Analysis (LDA), Multinomial Logistic Regression (MLR), Naive Bayes (NB), Decision Trees and Support Vector Machines (SVM) are six classifiers that are evaluated in this study. The results reveal the superiority of SVM Classifier in providing the best average accuracy.