Mixed logit model

This application uses full-information NNE to estimate a mixed logit model. The Matlab code is available in this GitHub repository. In this mixed logit model, the likelihood is easy to simulate, so full-information NNE does not really have an advantage over SMLE. But it offers a good setting to demonstrate and understand the method.

More details of this application can be found in our paper referenced on the Full-information NNE page.

Note

Scaffold page. This page mirrors the structure of the NNE consumer search code documentation. Fill in the example workflow and the file-by-file descriptions below with the contents of the mixed-logit repository (replace the placeholder GitHub link above with the real one).

An example to use the code

Run the scripts in order. This example is a Monte Carlo experiment that uses full-information NNE to estimate the mixed logit parameters from a simulated dataset.

>> monte_carlo_data        % simulate a dataset
>> fnne_gen                % generate the training examples (simulate datasets across theta draws)
>> fnne_train              % train the CNN and apply it to the data

Description of each file

model_mixed_logit.m

Describe the function that codes the mixed logit model (inputs, outputs).

fnne_gen.m

Describe how training examples \(\{\boldsymbol{\theta}^{(\ell)}, \mathcal{D}^{(\ell)}\}\) are generated.

fnne_train.m

Describe how the two-part CNN architecture is trained and applied to the data.