Time Series Energy Demand Forecast Using ANN Model Considering Various Influencing Meteorological and Economic Factors in Kano State, Nigeria

Abstract

Abstract This paper presents a model for forecasting daily energy demand in Kano State, Nigeria, using Artificial Neural Networks (ANN). The model addressed the challenge by leveraging historical energy consumption data together with weather variables (average temperature and humidity), interpolated population estimates and regional GDP, holiday indicators and temporal features. It was trained on data from January 2022 to December 2024 and tested on a held out 20 % set. Model performance was evaluated using the coefficient of determination (R 2 ), mean absolute error (MAE) and root mean square error (RMSE). The best ANN variant achieved an R 2 of 0.919, an MAE of 126.57 MWh and an RMSE of 207.15 MWh, demonstrating superior predictive accuracy compared to conventional time series methods. These results confirmed the robustness and practicality of the proposed approach for guiding energy planning in Northern Nigeria.