The first difficulty in using Statistical Analysis Software (SAS) Nonlinear (NLIN) models is that users have to provide the correct regression expression. The second difficulty to use SAS NLIN models is that the users have to provide the starting values of the parameters to be estimated. This is the first paper which tells all SAS users that artificial Higher Order Neural Networks (HONNs) can solve these two problems and HONNs are much better tools than SAS NLIN models for nonlinear analysis. This is the first paper which provides opportunities for millions of people working in the economics, accounting, finance and other business areas to know that the HONNs have more accurate application results, especially in the nonlinear data simulation and prediction area. HONN models are compared with SAS Nonlinear (NLIN) models and results show that HONN models are 3 to 12% better than SAS Nonlinear models.
展开▼
