A comparative study of time series analysis for forecasting energy demand in Nigeria
The complexity of the Nigerian Power Sector due to industry deregulation and abrupt variations in electricity demand and the ever increasing population density requires urgent research attention and industry action. Getting an accurate model to fit the energy demand pattern has become imperative and the need for an appropriate model cannot be overemphasized. This article therefore presents a comparative study on time series modelling of average and peak load forecasting in Nigeria. Data from the National Control Center (NCC), Oshogbo was harvested and analyzed using Harvey, Autoregressive, Moving Average and Exponential Smoothing Time Series Models while using R-Statistics for model simulation. The comparative performance showed that the Harvey Model best predicted load demand with least values of 111.0188 Root Mean Squared Error, (RMSE), 9.1095 Mean Absolute Error (MAE), 2.3579 Mean Absolute Percentage Error (MAPE) and 0.0156 Theil Inequality Coefficient (TIC) for average load and 117.4345 RMSE, 10.183 MAE, 17.01 Mean Absolute Percentage Error (MAPE), and 0.015 TIC for peak load.
Keywords: Forecasting, Electric load demand, Harvey model, Autoregressive model, Moving Average model, Time Series model, Exponential Smoothing model, Theil Inequality Coefficient