Verfication, Calibration and Validation in ABM

Verification:

• Focus: Ensures the model is built correctly and behaves as intended according to the programmed logic. It's an internal check to make sure the code translates the conceptual model accurately.
• Process: Verification involves techniques like code reviews, debugging, and tracing the execution of the model to identify and fix errors in the code itself.
• Example: Verifying if an agent always moves according to the programmed speed or if resource depletion is correctly calculated based on agent actions.

Calibration:

• Focus: Aligns the model's parameters with real-world data. It adjusts the model's behavior to match observed data from the system being studied.
• Process: Calibration involves adjusting model parameters (e.g., movement speed, resource consumption rates) to achieve a close resemblance between the model's output and real-world data. This might involve iterative adjustments and comparisons.
• Example: Calibrating the rate of sheep reproduction in a wolf-sheep model to match observed population growth rates in a real ecosystem.

Validation:

• Focus: Assesses whether the model accurately reflects the real-world system it represents. It checks if the model's predictions and behavior align with reality.
• Process: Validation involves comparing the model's output with data not used for calibration. This could involve comparing simulated outcomes with historical data, field experiments, or results from other established models.
• Example: Validating a traffic flow model by comparing simulated congestion patterns with actual traffic data collected from a specific road network.