Jacek Brożyna

Rzeszow University of Technology, Poland,

jacek.brozyna@prz.edu.pl

INFLUENCE OF DOUBLE SEASONALITY

ON ECONOMIC FORECASTS

ON THE EXAMPLE OF ENERGY DEMAND

Grzegorz Mentel

Rzeszow University of Technology, Poland,

gmentel@prz.edu.pl

Beata Szetela

Rzeszow University of Technology, Poland,

beata@prz.edu.pl

Abstract. The article deals with the issues of forecasting one of the most important elements of modern economies, which is electricity. The size of its production must be properly predicted and it should be technically possible and profitable from the economic point of view. The most frequently used forecasting models in literature are ARIMA and exponential smoothing. However, these models have a major disadvantage as they do not allow forecasting the series with double seasonality and periods of non-integer values. The authors of this article, on the example of data on maximum energy demand, presented an application of one of the latest forecasting models TBATS, which is devoid of this disadvantage. Using this model, the article presents forecasts for the year ahead with information on power consumption for each day, showing that long-term planning without losing details is possible. With such information energy producers are able to optimize production and the economic side of their business overall.

Received: March, 2016

1st Revision: June, 2016

Accepted: October, 2016

DOI: 10.14254/2071-8330.2016/9-3/1

JEL Classification: Q47, Q41, O13, N70

Keywords: electricity demand, forecasting time series with seasonality, TBATS, Fourier series, ARIMA, exponential smoothing