Time series are widely used for representation data of different types. Along the traditional methods the approach of nonlinear dynamics – reconstruction of the attractor of the system generating the series – is successfully applied. It allows us to calculate correlation dimension of the attractor of the system under study (if it exists) or to establish that the system does not have any attractor.
In this work we apply this method to solve a practical problem to analyze EEG records for revealing the patients with epileptic activity. Additionally we calculate entropy of a signal on amplitude coverage. This approach resulted in separation of 15 records into 2 classes – epileptic activity and other pathologies, and it is in accordance with the expert conclusions.
The implemented program system may be used both for investigations and educational purpose, and the method may be applied to time series of other types.