ApolloChoiceModelling forum

Dear all

I have two questions concerning the application of ‘apollo_prediction’ to MDCEV and mixed MDCEV models.

1. Following the instructions in the manual, I could get the predicted values (mean and standard deviation) for each alternative at the observation level when the input model was an MDCEV model. But, when I replaced the MDCEV model with a mixed MDCEV model (I added error components to the baseline utility function to the original MDCEV model) in the apollo_prediction command, I got only one value for each observation (the column head was ‘predictions’). How could I get the mean and standard deviation of the predicted values for each alternative in predicting a mixed MDCEV model? Part of the different prediction results is shown in the attached screenshots.

2. Instead of the summarized means and standard deviations, would it be possible to obtain the 200 simulated draws?

Any feedbacks are welcome. Thank you.

Regards,
Yishih

Hi Yishih

On point 1, this was a bug. This has been fixed in v0.1.1 which you can download from http://www.apollochoicemodelling.com/code.html already and which will be released on CRAN sometime this summer

On point 2, sorry, but these means and standard deviations are computed inside the code in an iterative manner and the values for the individual runs are not stored.

Best wishes

Stephane

Dear Dr. Hess:

Thank you for the prompt response. I was able to get the summarized statistics for each alternative by using the v0.0.1.

Regards,
Yishih