Fourier Residual Modified Approach in Group Method of Data Handling for Electricity Load Forecasting

Electricity is a highly essential part of daily life, including industrial, commercial, residential, and other sectors. A strong developed nation has significantly impacted the rise of electricity load demand. Hence, accurate load forecasting is fundamental for future planning in various sectors, enabling organizations to make informed decisions, optimize resource utilization, and adapt to changing demands in a dynamic environment. Hence, this paper proposes a residual modification by integrating a Fourier Series residual modification technique on the fitted Autoregressive Integrated Moving Average (ARIMA) time series model and an Artificial Intelligence (AI) Group Method of Data Handling (GMDH) model named F-ARIMA and F-GMDH respectively. The outcomes are compared with the conventional ARIMA and GMDH single model using the Mean Square Error (MSE) and Mean Absolute Percentage Error (MAPE). The results reveal that F-GMDH outperforms the model without employing Fourier residual modification.