Spatially Autocorrelated Residual Map

Model under- and overpredictions are clustered spatially.

Notice how the model underpredictions (red) are spatially clustered. Statistically significant spatial clustering of residuals (model over- and underpredictions) is evidence that your model is missing key explanatory variables. You can use the Spatial Autocorrelation tool to determine whether the spatial clustering of regression model residuals is statistically significant or not.