Add influencing variables to the upper level of nested logit model

[long pan]

Dear Prof. Bierlaire,

When I used the nested logit model example given in the website, the influencing variables are only included in the lower level of utility functions. I wondered if you could tell me how I can add variables to the upper level of my model?

For example, for the swissmetro problem, train and car option are in the existing nest, and SM is in future nest. I am planning to add variables to the existing utility instead of adding variables to train utility or car utility.

Thank you so much,

Long

[Bierlaire Michel]

On 11 Mar 2022, at 07:19, long pan <panlo...@gmail.com> wrote:

Dear Prof. Bierlaire,

When I used the nested logit model example given in the website, the influencing variables are only included in the lower level of utility functions.

Because it is how it should be. 

I wondered if you could tell me how I can add variables to the upper level of my model?

You should not. Just add the variables like if it was a logit model. The nested structure relates to the error term, not to the deterministic part of the utility. 

Otherwise, there are issues with the scale parameter that you do not want to deal with. 

For example, for the swissmetro problem, train and car option are in the existing nest, and SM is in future nest. I am planning to add variables to the existing utility instead of adding variables to train utility or car utility.

Thank you so much,

Long

--
You received this message because you are subscribed to the Google Groups "Biogeme" group.
To unsubscribe from this group and stop receiving emails from it, send an email to biogeme+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/biogeme/1f8f8736-46c0-4608-a658-32e4d02fc6ddn%40googlegroups.com.

[long pan]

Thank you so much for your reply!

Best regards,

Long