Sensitivity Analysis
- James W.
- 3 days ago
- 1 min read
LinkedIn Post 13: Sensitivity Analysis
Your synthetic comparable algorithm produces a value estimate. But what if you changed an input? What if the property were 10% larger? 5 years older? In better condition?
Sensitivity analysis tests whether outputs respond appropriately to input changes. Synthetic comparables should show stable, economically reasonable sensitivity.
If small input changes cause massive value swings, the synthetic comparable is unstable. If the algorithm is insensitive to inputs that should matter significantly (property age, condition), the model is suspect.
Run sensitivity analysis before relying on synthetic comparables. Document results. Include in validation record.
Understanding model sensitivity is essential to trusting model output.
Comments