Comparison Between Ordinary Methods (LS,IV) and Robust Methods (2SWLS,LTS,RA) to estimate the Parameters of ARX(1,1,1) Model for Electric Loads

Abstract

The models of time series often suffer from the problem of the existence of outliers that accompany the data collection process for many reasons, their existence may have a significant impact on the estimation of the parameters of the studied model. Access to highly efficient estimators is one of the most important stages of statistical analysis, And it is therefore important to choose the appropriate methods to obtain good estimators. The aim of this research is to compare the ordinary estimators and the robust estimators of the estimation of the parameters of the Autoregressive with exogenous variable (ARX) model with the order of (1,1,1) using real data containing outliers, the order (1,1,1) has been used based on a number of criteria for determining the rank, which were explained in the thesis under construction. The study showed that the method employed The Least Trimmed Squares (LTS) method is the best method of estimation. The comparison was done using the Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) and Expected Error Percentag (EEP), A test was also carried out to ascertain the accuracy of the model reached and then used to predict future values.