Time Domain Analysis
In the frequency domain discussed in the previous chapter, a spectral approximation is used for an arbitrary second-order stationary time series ..., y -1, y 0, y 1,.... In the traditional approach, the variance components in the spectral approximation will be nonzero, thus there is little chance of developing a parsimonious model. (The model with measurement error is more promising in this regard.)
KeywordsAkaike Information Criterion Kalman Filter Covariance Function ARIMA Model Time Domain Analysis
Unable to display preview. Download preview PDF.