Accurate power forecasting for electrical consumption prediction is crucial for national energy planning since it is a method for understanding and anticipating market energy demand. In a deregulated market, their power output can be modified accordingly. In order to explicitly deal with seasonality as a class of time-series forecasting models, Persistence Models (Naive Models), Seasonal Autoregressive Integrated Moving Averages with Exogenous regressors, and Univariate Long-Short Term Memory Neural Network (LSTM) are utilized. The main goal of this project is to do an exploratory data analysis of the major nations' power systems, followed by the application of several forecasting models to determine the daily peak demand and the next 24 hours' worth of energy need once every day. Energy is a vital resource for society's development. The development of the economy depends heavily on an accurate assessment of its consumption. A database of historical energy consumption statistics from all the main nations of the globe was created. The complete information was modelled and simulated using machine learning techniques to anticipate the energy consumption of each country because national patterns may be transferrable from one country to another. Understanding the commonalities between the characteristics of other nations might boost prediction accuracy and enhance the corresponding public policies.
Keywords
LSTM, Autoregression, World Countries, Energy Consumption
