Best Estimation Of Double Seasonal Pattern Electric Load Parameters Using Least Squares Method

Abstract

Forecasting is an important tool in planning an effective and efficient use of electrical loads. This paper presents an improvement in parameter estimation from previous studies. The results of previous studies indicate that the DSARIMA model is with MAPE about 2.06 percent. This model produces white noise residuals, but not normally distributed, which is thought to be due to outliers. Data smoothing is done to get the best data pattern. The analysis results show that the AR parameter iteration of the best DSARIMA model that is appropriate for short-term forecasting is with MAPE about 1.56 percent.