Counterfactuals

Important Links

• CCP Overview

• Quick Start

• Simulation Study

• Bus Engine Example

Use the public counterfactual method to solve a fitted CCP model under new reward parameters or a replacement transition tensor.

cf = model.counterfactual(replacement_cost=4.0)

print(f"replacement_cost={cf.params['replacement_cost']:.6f}")
print(f"P(replace | state=50)={cf.policy[50, 1]:.6f}")

Result

replacement_cost=4.000000
P(replace | state=50)=0.054908

The method returns a parameter dictionary, value function, and policy. Pass transitions=new_tensor to change the transition law. A single call can change reward parameters or transitions, but not both. The tensor orientation is (n_actions, n_states, n_states). Removing an action requires a model with a reduced action space.

Results

The 100-state study evaluates reward and transition changes over 20 independent panels:

Change	Mean policy TV	Mean value loss
Increase the first reward parameter by 1.0	0.0022	0.000184
Slow deterioration	0.0019	0.000083

Policy TV is the state-averaged total-variation distance between the fitted counterfactual policy and the policy from the true parameters. Value loss measures the cost of using the fitted policy instead of the true-parameter policy. See the Simulation Study for the estimation and inference results.