As the global shift towards sustainable energy resources continues, the forecasting accuracy of electricity demand becomes increasingly important for utilities and grid operators. Accurate predictions enable better planning for power production and distribution, as well as optimization of pricing and balancing of supply and demand.

Electricity demand forecasting is a complex task that can be impacted by a wide range of factors such as weather, electricity prices, and consumer behaviors. To address this challenge, a variety of methods have been employed, including statistical methods and machine learning algorithms, with an increasing focus on deep learning models as the amount of available data and computational power has grown.

However, the increasing complexity in the energy sector with the integration of distributed energy resources and various technologies introduces new challenges for load forecasting. Furthermore, the specific purpose of the forecast, whether it be for short-term, long-term, or peak load forecasting, also presents unique challenges. Moreover, different sources of electricity load, such as private households, industries, or office buildings, require different supporting data and present additional conditions that further complicate the accuracy of an electricity load forecast.