Modeling Method and Algorithm

The default modeling method is NARX Structure, the default modeling algorithm is ASCMO Gaussian Process Spectrum (ASC GP-Spectrum), which represents the recent improvement of the standard ASC algorithm to cope with large data sets.

ASC GP-Spectrum requires a number of Basis Functions s set by the user. During the model training, the information contained within the entire data set will be transformed to a set of s virtual basis points. The bigger the size s, the better the modeling result, but this goes along with an increase in modeling time. The recommended range is 50 < s < 200.

The number of Basis Functions, as well as other model parameters, are set in the <output> - Parameters window. This window opens automatically when you select the model type ASC GP-Spectrum; in addition, you can open it with the Edit button.

The ASCMO Gaussian Process Sparse Constant Sigma (ASC GP-SCS) model type is recommended for large numbers of training data.

The ASCMO Gaussian Process (ASC GP) model type is recommended for small numbers of training data.

A Linear model type is provided to train a linear model. For the linear model, the <output> - Parameters window offers fewer parameters to edit.

Advanced Parameters for ASC GP Models can be set by using the Edit button, Advanced Settings must be activated via File > Options.

For more information on modeling methods and algorithms, see the online help (<F1>).