# Notes for reviewing Time series.docx - Notes for reviewing...

• 3

This preview shows page 1 - 2 out of 3 pages.

Notes for reviewing Time series 1. Residual mean square error: s 2 = t = 1 n e t 2 n r = t = 1 n ( Y t ^ Y t ) 2 n r ( 1 ) e t : the residualat time t n: number of residuals r: the total of parameters estimated n-r: degree of freedom. Formula (1) is the same as : s 2 = t = 1 n ( Y t ^ Y t ) 2 n k 1 = SSE n k 1 = MSE ( 2 ) n : number of observations k: number of independent variables in the regression function MSE = SSE n k 1 : residual mean square error n-k-1: degree of freedom From (1) and (2), we can see that: r= k-1 2. You should virtually never have to use a value of p or q larger than 3 in an ARIMA model for business application , and in most cases they are less than 3 . Also, the sum of p and q will generally be no larger than 3, and usually only one of the two will be non-zero . You should try to avoid using “mixed” models in which there are both AR and MA coefficients, except in very special cases. 3. Implementing the model building strategy - Model identification: o Define whether the series is stationary. If not, make
• • • 