Electricity Consumption Forecasting in Thailand using Hybrid Model SARIMA and Gaussian Process with Combine Kernel Function Technique

Abstract

Electricity consumption forecasting plays a significant role in planning electric systems. However, this can only be achieved if the demand is accurate estimation .This research, different forecasting methods hybrid SARIMA-ANN and hybrid model by SARIMA- Gaussian Processes with combine Kernel Function technique were utilized to formulate forecasting models of the electricity consumption . The objective was to compare the performance of two approaches and the empirical data used in this study was the historical data regarding the electricity consumption (gross domestic product: GDP, forecast values calculated by SARIMA model and electricity consumption) in Thailand from 2005 to 2015. New Kernel Function design techniques for forecasting under Gaussian processes are presented in sum and product formats. The results showed that the hybrid model by SARIMA - Gaussian Processes with combine Kernel Function technique outperformed the SARIMA-ANN model have the Mean absolute percentage error is 4.7072e-09, 4.8623 respectively.