Int. J Sup. Chain. Mgt Vol. 8, No.5, October 2019 467 Forecasting of Electricity Consumption and Supply for Campus University using Time Series Models Rosnalini Mansor #1 , Bahtiar Jamili Zaini #2 , Chong Shi Yee *3 # School of Quantitative Sciences, Universiti Utara Malaysia, 06010 UUM Sintok Kedah Malaysia * Gamuda Land (T12) Sdn. Bhd., 47820 Petaling Jaya, Selangor, Malaysia 1 rosnalini@uum.edu.my 2 bahtiar@uum.edu.my 3 sychong92@hotmail.com Abstract— Electricity is an important energy source in university as lecture classes need electricity supply to function. It is also important for the development of the university. Since electricity consumption is a necessity of a university’s operation, the forecast of electricity consumption on the university campus should be made. This is essential for the development of the university as the treasury department can manage the funding from the government according to the value forecasted to make full use of the funding in the university’s development. There are several forecasting methods used in this study, including time series regression, seasonal exponential smoothing, Box-Jenkins (SARIMA), decomposition and the naïve method. Error measurements used to evaluate the performance of forecasting model were mean square error (MSE), root mean square error (RMSE), mean absolute percentage error (MAPE) and geometric root mean square error (GRMSE). The results of this study showed that the seasonal exponential smoothing model was the best in the 1-step ahead and 2-step ahead forecasting while SARIMA (0,2,2)(0,2,1)12 was the best in the 3-step ahead forecast. The overall performance of seasonal exponential smoothing was the best in this study. Throughout this study, suggestions were made for the next study regarding electricity consumption in university to consider factors such as semester breaks and students’ activities in order to examine its effect in electricity consumption. Keywords— Forecasting, Electricity Consumption, Univariate Time Series, Box-Jenkins 1. Introduction Electrical consumption is the total amount of energy used represented by kilowatt hours (kWh). It is different from load demand which means the immediate rate of that consumption (kW). For example, for a light bulb using 100 watts of electricity that is switched on for 10 consecutive hours, the consumption is 1kWh. Alternatively, ten 100 watts light bulbs switched on at the same time for an hour has the same consumption (1kWh), but its load demand is 1kW of electricity to operate. The forecast electricity consumption in terms of kWh is due to the policy in Malaysia. Electricity tariffs will change, therefore, forecasting based on electricity charges in Ringgit Malaysia (RM) is not meaningful. Electricity consumption (kWh) is more representative as it shows the actual usage every month without the influence of electricity tariff. Electricity is an important energy source in each country. It is also important for the development of a university. Other than basic facilities such as lecture halls, student residential halls, and the library, a university also provides facilities such as the sports center, health center, food court, smart reading room, gym room, swimming pool, and many more. Without the supply of electricity, electronic components such as lights, fans, air-conditioners, projectors and computers cannot function. This will affect the development and learning process of students in that campus. The monthly expense for the electricity bill is different as the activities in the university would affect it. Electricity tariff also changes from time to time, thus, studies regarding electricity consumption are more representative. Therefore, our objective for this study is to evaluate the performance of several forecasting models by using the four error measures to then forecast the monthly electricity consumption. 2. Literature Review Articles related to electricity consumption in Malaysia and foreign countries were studied. Researchers used different forecasting methods in their study according to data type and regional factors. Chujai, Kerdprasop and Kerdprasop [1] forecasted the electricity consumption in an individual household by performing the Autoregressive Integrated Moving Average (ARIMA) and Autoregressive Moving Average (ARMA) model. The data was from December 2006 to November 2010. The result shows that ARIMA is suitable for monthly and quarterly forecasting, while ARMA is suitable for daily and weekly forecasting. Besides that, Box-Jenkins Seasonal Autoregressive Integrated Moving Average (SARIMA) was carried out to forecast electricity consumption in Malaysia [2]. ______________________________________________________________ International Journal of Supply Chain Management IJSCM, ISSN: 2050-7399 (Online), 2051-3771 (Print) Copyright © ExcelingTech Pub, UK (http://excelingtech.co.uk/)