PERBANDINGAN METODE VECTOR ERROR CORRECTION MODEL (VECM), VECTOR AUTOREGRESSIVE (VAR), DAN FUNGSI TRANSFER

The initial process of forecasting methods such as: Vector Error Correction Model (VECM), Vector Autoregressive (VAR), and Transfer Function has some similarities like: determining the optimal lag and Granger Causal testing of the variables to be modeled. Sometimes the mathematical models and forecasting results produced significantly different. The purpose of this study was to compare methods : VECM, VAR, and Transfer Function in modeling the relationship between the data of water flow and rainfall. These data,
obtained from the measurements made by the administrator of Hydroelectric Power Plant (PLTA) Tes in
Lebong Province of Bengkulu. The method used to achieve these goals : starting with data processing for each
forecasting method to obtain the best mathematical models of each method. To find the best model of the three
models were produced, used the smallest Mean Square Error (MSE). To simplify the processing of the data, we
use softwares SAS and Minitab. The results showed that the value of each MSE from VECM, VAR, and Transfer
Function successively: 90.40; 85.53; and 86.88. Based on the results, the best method for a case study in this
research is the method of Vector Autoregressive (VAR) with MSE 85.53. The mathematical models are as
follows :