| System Identification Toolbox™ | |
m.EstimationInfo m.es m.es.DataLength, etc
Any estimated model has the property EstimationInfo, which is a structure whose fields give information about the results of the estimation. Depending on whether it is an estimated parametric idmodel or an estimated frequency response idfrd, EstimationInfo will contain different fields.
The model structure will contain the properties ParameterVector, CovarianceMatrix, and NoiseVariance, which are all calculated in the estimation process (see the reference page for idmodel). In addition, EstimationInfo contains the following fields:
Status: Information whether the model has been estimated, or modified after being estimated.
Method: Name of the estimation command that produced the model.
LossFcn: Value of the identification criterion at the estimate. Normally equal to the determinant of the covariance matrix of the prediction errors, that is, the determinant of NoiseVariance. Note that the loss function for the minimization might be different due to LimitError. The value of the nonrobustified loss function is always stored in LossFcn.
FPE: Akaike's Final Prediction Error, defined as LossFcn *(1+d/N}/(1-d/N), where d is the number of estimated parameters and N is the length of the data record.
DataName: Name of the data set from which the model was estimated. This is equal to the property name of the iddata object. If this was not defined, the name of the iddata variable is used.
DataLength: Length of the data record.
DataTs: Sampling interval of the data.
DataDomain: 'Time' or 'Frequency', depending on the data domain.
DataInterSample: Intersample behavior of the data from which the model was estimated. This equals the property InterSample of the iddata object. (See iddata.)
WhyStop: For models that have been estimated by iterative search. The stopping rule that caused the iterations to terminate. Assumes values such as'MaxIter reached', 'No improvement possible along the search vector', or 'Near (local) minimum'. The latter means that the expected improvement is less than Tolerance (see Algorithm Properties).
UpdateNorm: Norm of the Gauss-Newton vector at the last iteration.
LastImprovement: Relative improvement of the criterion value at the last iteration.
Iterations: Number of iterations used in the search.
InitialState: Option actually used when Model.InitialState = 'auto'.
N4Weight: For n4sid estimates, or estimates that have been initialized by n4sid: the actual value of N4Weight used.
N4Horizon: For n4sid estimates, or estimates that have been initialized by n4sid: the actual value of N4Horizon used. See n4sid and Algorithm Properties.
If the idfrd model is obtained from an estimated parametric model,
g = idfrd(Model)
g.EstimationInfo is the same as Model.EstimationInfo as described above.
For an idfrd model that has been estimated from etfe, spa, or spafdr, EstimationInfo contains the following fields:
Status: Whether the model is estimated or directly constructed.
Method: etfe, spa, or spafdr
WindowSize: Resolution parameter (or vector) used for the estimation
DataName, DataLength, DataTs, DataDomain, DataInterSample: Properties of the estimation data as above.
| Algorithm Properties | |
| idpoly | |
| idss |
| d2c | etfe | |
| © 1984-2008- The MathWorks, Inc. - Site Help - Patents - Trademarks - Privacy Policy - Preventing Piracy - RSS |