Epileptic Seizure Prediction for Imbalanced Datasets

In this study, the methods used in the classification of imbalanced data sets were applied to EEG signals obtained from epilepsy patients and epileptic seizures were estimated. Firstly, the data set was balanced by using under-sampling, over-sampling, and synthetic minority over-sampling technique and classified with Support Vector Machines. Then, the data set was classified using the Rusboost classifier without balancing. Classification results were compared with different criteria and the advantages and disadvantage of the methods were evaluated.