Özet:

In this study, for the period between January 2016 and December 2018, forecast models on electricity consumption data in Turkey were created. The forecast approach was used as "Fourier Analysis with the Smallest Square Method" and "Winters' Method". The Smallest Quadratic Method and Fourier Analysis used Microsoft Excel for model equation and the Winters Method for modeling with the Minitab 19 Package Program. In the predictive modeling, electricity consumption values (TWh) are periodically studied, variability (variation) ratio calculated for 12 months, 24 months and 36 months. In this context, the periodic variability of the data set and the success of the forecast models; the variability rate (Cv) is assessed by MAPE, MSE, RMSE and MSD success criteria. In this study, the success of modeling for the observation data of time series containing seasonal volatility such as electricity consumption has been compared to high alternative predictions models. The results of the evaluation are estimated by the Winters’ Method, which provides better results in terms of success criteria for the next 36 months of electricity consumption. Created models are a guide for strategy determiners.

Anahtar Kelimeler: