Throughout the last decades there has been an increasing interest in analyzing the EEG signals of epilepsy patients in order to relate it to epilepsy seizure onsets. Previous research papers were published exploring the possible techniques to utilize the EEG signals for detecting and predicting seizure onsets through Machine Learning and Deep Learning models, such as Support Vector Machines and Convolutional Neural Networks. The aim of this work is to build practical hardware-implementable Machine Learning classifiers capable of predicting the seizure onsets prior to their occurrences with high sensitivity and accuracy. The classification method proposed involves removing certain channels for each patient, extracting the features from the EEG signal, selecting the best feature combination for each patient, and finally training the selected SVM classifier accordingly. Evaluating the performance of the proposed classification technique yields promising results for the selected patients with accuracies exceeding 95%.