STAT443 Lecture Notes - Lecture 2: Autoregressive Conditional Heteroskedasticity, Xm Satellite Radio, Time Series
Document Summary
Recall from fitting a gaussian process to a time series xi xa. Assuming stationary are reduced the parameters from n ninth. For example we assume follow an mac"s model. Xt ut ozc. it ze where ze d nco 021. Or we assume xi follow an arg model with 3 parameter cu 0 021. An auto regression model of order psi having the representative of xt it pixel 02xt2 where 124 is a white noise process with mean q it 0pxep 120 variance or is defined as a stationary process. The parameters in 01 1oz cfp ont are unknown ol needed to be estimated. Note that in archmodel the current xt is represented through its immediate. Xt p in a linear regression form an hrcp p past values xel xt2 is easy to implement and therefore is the most popular time series model in practice for p i ca first order autoregressive xt ut qxt.