PREDICTION OF ELECTRICITY CONSUMPTION USING MLR

Multiple linear regression (MLR) method was applied in this research to predict electricity consumption in Chonburi province, Thailand, using three independent variables; number of electricity customers, rainfall and temperature to fit the equation. All data was separated into two data sets; (1) training data set for generating the prediction equation and (2) validation data set for validating the accuracy performance by mean absolute percentage error (MAPE). All assumptions of MRL analysis were tested by Anderson-Darling statistic (AD), Durbin-Watson statistic (DW), Breusch-Pagan statistic (BP) and Variance inflation factor (VIF). Finally, the prediction performance was illustrated by the regression standard error, the adjusted coefficients of determination and MAPE with 98,453 kWh, 0.928 and 4.65%, respectively.