After building a bayesian model it would be great to have simple way to run counterfactual analysis. Perhaps a procedure where we could specify the model and a variable that we want to simulate different values for. The output could include the target variable values for the different input variable values and a sensitivity (tornado) diagram.