choice-metrics.com

Thank you prof. Bliemer and sorry for my blunder in posting question in another thread. I am again following you here, so that we won't get confused. I have been trying to run this model but it has been super slow. Can you please check , if there is any problem in the model? Again, thank you very much.

Pramisha

It is slow because for each design evaluation it is doing 3^10 = 59,000 draws, and the modified Federov algorithm needs to cycle through all (2000) candidates, which is also very slow.

Since you are not using reject or require constraints, there is no real need to use the modified Federov algorithm, so I suggest that you use the default swapping algorithm.

Further, you either need to reduce the number of Bayesian priors (you can make the Tillage and Cover Crops coefficients fixed because they are the least important attributes in terms of contribution to utility) or you can restrict the number of draws.

My suggested syntax in below.

Code: Select all

Design
;alts= optA*, optB*,Conventional Tillage
;rows=36
;eff=(mnl,d, mean)
;block=6
;bdraws=sobol(5000)
;model:
U(optA) = b1 [(n, 0.16377, 0.01022)]                                           * Carbon_Payment[0,5,10]
        + b2.effects[(n, 0.26093, 0.05064)]                                    * Tillage[1,0]
        + b3.dummy[(n, 0.26093, 0.05064)]                                      * Cover_Crops[1,0]
        + b4.dummy[(n,-0.20278, 0.04824)|(n,-2.0, 0.09)|(n,-1.04265, 0.06913)] * Certification[3,2,1,0]
        + b5.effects[(n, -0.86152, 0.05450)]                                   * Contract[1,0]
        + b6.dummy[(n,-0.20278, 0.04824) | [(n,-1.04265, 0.06913)]             * Source[2,1,0] /
U(optB) = b1 * Carbon_Payment
        + b2 * Tillage
        + b3 * Cover_Crops
        + b4 * Certification
        + b5 * Contract
        + b6 * Source /
U(Conventional Tillage) = b0 [(n, 0.77716, 0.18272))]
$

Michiel