Multiple equations model selection algorithm with iterative estimation method

A good model is a model that encapsulates the initial process and therefore represents a close estimate to the true model that generated the data.However, whenever there is more than one model to be considered, selection decision needs to be based on its competence to generalize, which is defined as a model’s ability to fit not only current data but also to forecast future data. There have been various procedures suggested to date, whether through manual or automated selections, to choose the best model.This study nonetheless focuses on an automated selection for multiple equations model with the use of iterative estimation method. In particular, an algorithm on model selection for seemingly unrelated regression equations model using iterative feasible generalized least squares estimation method is proposed. This estimation method is equivalent to maximum likelihood estimation at convergence. Therefore, the algorithm is known as SUREIFGLS-Autometrics.Simulation and real data analyses were conducted to assess the performance of the algorithm.The simulation results reveal that the algorithm shows almost similar performances for all conditions involved. Meanwhile, real data analysis using water quality index displays excellent accomplishments when compared to other selection procedures.Consequently, iterative feasible generalized least squares method is regarded as a more suitable estimation method in this automated selection.It can also be seen that simultaneous selections outperform the individual selections.This strategy by executing simultaneous selection with iterative estimation method is therefore proven to outclass in this analysis