Model Validation using Biogeme

257 views
Skip to first unread message

Ishant Sharma

unread,
Mar 18, 2019, 2:52:16 PM3/18/19
to Biogeme
Dear Prof Bierlaire,

I was wondering if PandasBiogeme supports holdout (train and test data split) model validation for any choice model.

Presently I am estimating an ICLV model with 70% data (training data). I have no idea how to use the remaining 30% data to test the model while using the estimated model to predict the outcomes in PandasBiogeme. 

Please help or direct me further to the applicable documentation if it is possible to do so. 

Thank you

Ishant 

Michel Bierlaire

unread,
Mar 18, 2019, 3:26:49 PM3/18/19
to reachish...@gmail.com, Michel Bierlaire, Biogeme

On 18 Mar 2019, at 18:38, Ishant Sharma <reachish...@gmail.com> wrote:

Dear Prof Bierlaire,

I was wondering if PandasBiogeme supports holdout (train and test data split) model validation for any choice model.

Presently I am estimating an ICLV model with 70% data (training data). I have no idea how to use the remaining 30% data to test the model while using the estimated model to predict the outcomes in PandasBiogeme.

As PandasBiogeme is a Python package, you can do whatever you want. It is indeed easy to do what you describe, but you have to program it. 
Look at the examples. Basically, you  need to combine the examples
and

or

and


 

Please help or direct me further to the applicable documentation if it is possible to do so. 

Thank you

Ishant 

--
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 post to this group, send email to bio...@googlegroups.com.
Visit this group at https://groups.google.com/group/biogeme.
For more options, visit https://groups.google.com/d/optout.

Reply all
Reply to author
Forward
0 new messages