The plots show a residual analysis of the selected models.
For each of the models, the residuals (prediction errors) are computed for the data selected as the Validation Data set.
For Time Domain Validation Data, the autocorrelation function for the residuals as well as the cross correlation function between input and residuals are computed and displayed. The confidence interval for these functions is shown by default as dashed lines, and can be turned off using the appropriate Options submenu. Ideally for an acceptable model, the correlation curves should lie between these lines.
For Frequency Domain Validation Data, the estimated power spectrum of the residuals is shown, as well as amplitude plot of the transfer function from input to the residuals. Also in this case the upper and lower confidence intervals of these estimates are shown as dashed lines. The interpretation is that for the transfer function estimate, the confidence interval should contain zero (i.e. the lower confidence line should be "below zero") to indicate an acceptable model. Frequency regions where this is not the case indicate regions where the model apparently not have captured all information in the data. Correspondingly, for white residuals, it should be possible to draw a straight line within the confidence interval or the power spectral plot. (For IDFRD validation data, the confidence intervals may be misleading, depending on the amount of data compression in the IDFRD data set.)
(file resid.htm)