PREDICTION OF ELECTRICAL ENERGY DEMAND USING THE MULTIPLE LINEAR REGRESSION METHOD: CASE STUDY OF THE CITY OF NDJAMENA

The need of electrical energy has been increasing lately in developing countries. However, several methods for predicting the electric charge exist: such as statistics, artificial intelligence and hybrid approaches. This work focuses on the modeling of electrical energy demand by the multiple linear regression method in the case of the National Electricity Society (NES) of the city of NDjamena. The estimates obtained are based on statistical analyses carried out on 5 exogenous variables. The results of the analyses gave very good meanings through the values of the standard errors associated with the regression coefficients. The two configurations developed all have average absolute MAPE errors of less than 2%. In the first configuration, we obtained an adjusted R² coefficient of determination of 0.975, a standard error of 30.395 GWh and an RMSE of 10.1 GWh. While the second configuration gave an R² (adjusted) equal to 0.974, with a standard error of 31.092 GWh and an RMSE of 10.28 GWh. The latter is made up of (3) parameters validated by the statistical indicators of the step-by-step downward regression. All of our results have shown that with this method, we can estimate an adequacy of 951 GWh to meet electricity need by 2035. The purpose of this study is to recommend a simple and efficient model for the prediction of the electric charge.