Exponential smoothing is among the readily used class of methods for smoothing a discrete time series for the purpose of forecasting the future state. Its computational efficiency, simplicity and ease to adjust its responsiveness to process variations to be forecasted have contributed to its popularity (Ostertagová & Ostertag, 2011). The main idea of the procedure is to smooth the original time series in the same way as moving average and using the smoothed series to forecast future value of required variable. However, in exponential smoothing the more recent values of the time series are allowed to have huge influence on the future value forecast than the older observations. The exponential smoothing approach is pragmatic and simple in forecasting in which the forecasted value is built from exponentially weighted average of more recent observations with the less weight given to most recent observation while largest weight to the present observation.
