Covariance Parameters

Figure 1. Estimates of covariance parameters for autoregressive covariance model
Estimates of covariance parameters for autoregressive covariance model

Both of the residual covariance parameters are statistically significant. The negative AR1 rho suggests that high sales one week is is associated with lower sales the next.

The random effects variance estimate is also statistically significant. Its value is larger than the residual variance (the AR1 diagonal), suggesting that most of the variability unaccounted for by the fixed effects is due to market-to-market sales variation.

Next