Epilepsy is a neurological condition associated with recurrent seizure. It is a result of excessive and abnormal neuronal firing, the exact cause of which is unknown. Diagnosis often involves ruling out other conditions with similar symptoms, however electroencephalogram (EEG) can help to confirm the exact cause. The overall process is lengthy, time consuming and needs expertise separating neuronal signals from EEG artefacts to aid in proper diagnosis. Further, it was observed that clinicians encounter EEG signals where they face difficulty in classifying them as either normal or epilepsy; we term these as intermediate EEG signals. These recordings create a lot of confusion in clinical decision-making thus delaying the diagnosis and therefore affect treatment. Clinicians are usually inclined towards further observations and patients are treated as and when required. Hence, automated analysis and correct classification of EEG data with the help of computational techniques can be helpful in interpreting such recordings as either normal or epilepsy. Here, we will discuss on different feature engineering and machine learning techniques to identify signatures and rules that help to differentiate epilepsy patients from normal individuals. Some of these signatures were found to be present in patients with epilepsy while absent in normal individuals, and vice versa. Thus, these identified class signatures instil confidence to accurately detect patients with epilepsy and help in complex decision-making process.