How does uncertainty differ from model performance evaluation using statistical metrics such as MSE, RMSE, MAE, MAPE, and R² in the field of real estate appraisal?
ALL of the measures cited are just mathematical calculations that indicate a particular kind of variation, or uncertainty, in a data set. There is no specific rule on which one you should use apart from the constraints (e.g. linearity) of some of the methods.
Uncertainty in extrapolations degrades quickly with future time. These calculations may be part of the assessment of a real estate appraisal methodology, but they are not directly comparable to the complete "Model Performance Evaluation" (unless you define it as the sole criterion).
While statistical metrics are available to measure model performance from current data, uncertainty is left without tools to assess the model's future accuracy.