The neurological condition known as epilepsy, which is marked by recurring seizures, presents considerable obstacles to prompt diagnosis and treatment. We investigate the effectiveness of hierarchical clustering (HC) as a feature extraction technique for pre-processing electroencephalogram (EEG) signals using data from publicly accessible datasets. Preictal and ictal state-balanced samples are included in the dataset, enabling reliable model training and assessment. The performance of hierarchical clustering as a pre-processing technique for a biological signal like EEG is evaluated by training couple of models and obtaining the evaluation metrics from them and finally all the metrics are compared together.