Time-series trendline and curve-fitting-based approach to short-term electricity demand forecasting

Electricity load demand forecasting and its accuracy is an important process for utility planning, maintenance, scheduling, operation, and control in power systems. Historical data are also very vital in demand forecasting processes. This study examined weekly electricity demand forecasting model using trendline methods which include linear trendline, moving average, exponential smoothing, quadratic, and logarithmic trends. The calculations and analysis were carried out using Microsoft Excel. The results were compared using known performance evaluation metrics such as mean absolute percentage error (MAPE) and root mean square error (RMSE). Cubic root mean error (CRME) was introduced as a performance evaluation metric. The hybrid (quadratic-logarithmic) method was found to outperform the other individual trendline methods. This method produced the lowest value of MAPE, RMSE, and CRME representing 14.41%, 14.68%, and 14.65% respectively which indicated that hybrid model performs better than individual models operating separately when used in forecasting.