260. VARIMAX
Vector ARIMA with exogenous regressors
= endogenous vector series
= exogenous regressor vector
= matrix of regression coefficients
Each of the series can depend on the same set of exogenous variables.
Parameters: , , (), ,
Orders: , , , , (regressors)
Example:
Given
- Orders: , , , , (one exog)
-
Coefficient matrices:
- Intercept:
- Initial conditions: ,
- Endogenous :
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | |
| 12 | 10 | 8 | 11 | 14 | 12 | 9 | 13 | 16 | 14 | 11 | 15 | 18 | 16 | 13 | 17 |
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | |
| 8 | 9 | 7 | 6 | 10 | 11 | 9 | 8 | 12 | 13 | 11 | 10 | 14 | 15 | 13 | 12 |
- Exogenous :
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | |
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 |
Step 1 — formula
Substitute orders into the VARIMAX recursion (with , no differencing):
Forecast (set ):
Three contributions: AR product, exogenous regression , MA product.
Componentwise:
Innovation:
Step 2 — apply at
Compute the three contributions separately, then sum.
AR part:
Exog part:
MA part: . With , the step gives and .
Sum:
Step 3 — iterate
Three contributions per row: AR , exog , MA . Values rounded to 4 decimal places.
| 1 | |||||
| 2 | |||||
| 3 | |||||
| 4 | |||||
| 5 | |||||
| 6 | |||||
| 7 | |||||
| 8 | |||||
| 9 | |||||
| 10 | |||||
| 11 | |||||
| 12 | |||||
| 13 | |||||
| 14 | |||||
| 15 | |||||
| 16 |