Combining multiple linear univariate features in one feature space and classifying this feature space using machine learning methods could predict epileptic seizures, in the patients suffering from refractory epilepsy. For each patient, a set of twenty-two linear univariate features are extracted from 6 electroencephalogram (EEG) signals, and make a 132 dimensional feature space. Preprocessing and normalization methods of the features, affect the outputs of the seizure prediction algorithm, and are studied in terms of alarm sensitivity and false prediction rate (FPR). The problem of approximately choosing an optimal preictal time was tackled using 4 distinct values; 10, 20, 30, and 40minutes Seizure prediction problem has traditionally been considered as a two-class classification problem, which is also exercised here. The studies have been done on the features obtained from 10 patients. For each patient, 48 different combinations of methods are compared to find the best configuration. Normalization through dividing by maximum and smoothing, are found as the best configuration in most of the patients. Results also indicate that applying machine learning methods on a multidimensional feature space resulting from 22 univariate features could predict seizure onsets with high performance. In average, the seizures were predicted in 73.9% of the cases (34 out of 46 in 737.9hours of test data), with a FPR of 0.15h(-1).