Page 1 of 1

Some variables do not influence the log-likelihood of your model!

Posted: 01 Apr 2022, 13:20
by Hyewon
Dear Stephane,

Hello, this is Hyewon.
I know there are several questions about this error message, but I cannot find the way same as mine and I think I code in the right way.

Currently, I am using Apollo version 0.2.7.
I am using ICLV.
The choice part is the binary mixed logit model.
I have four latent variables which are named operation, regulation, rideshare, and sustain.

The latent variables named with the operation, regulation, and rideshare are on a 5-point Likert scale.
The "sustain" is a binary variable with 4 indicators.

The error message that I get is,

Testing influence of parametersError in apollo_estimate(apollo_beta, apollo_fixed, apollo_probabilities, :
Parameters zeta_operation_1, zeta_operation_2, zeta_regulation_1, zeta_regulation_2, zeta_regulation_3, zeta_rideshare_1, zeta_rideshare_2, zeta_rideshare_3, zeta_rideshare_4, zeta_sustain_1, zeta_sustain_2, zeta_sustain_3, zeta_sustain_4, sigma_operation_1, sigma_operation_2, sigma_rideshare_1, sigma_rideshare_2, sigma_rideshare_3, sigma_rideshare_4, sigma_regulation_1, sigma_regulation_2, sigma_regulation_3 do not influence the log-likelihood of your model!

===============

I think the model cannot understand the whole latent variable measurement component, I don't know why..
Can you please take a look and see what is the problem with it?

Since the code is long and so easy to distinguish, here I attach the gist link of GitHub.


https://gist.github.com/HyewonNamgung/a ... dadb32ff55

Thank you so much always for your effort in this wonderful tool.

Re: Some variables do not influence the log-likelihood of your model!

Posted: 04 Apr 2022, 17:22
by stephanehess
Hi

this one was easy to fix.

You're not actually using those model components. You have created P[["indi_operation_1"]] etc, but then later on, you have:

P[['model']] = apollo_mnl(mnl_settings, functionality)

this means that you are only using the mnl part.

What you need to use is something like P[['choice']] = apollo_mnl(mnl_settings, functionality)

and then

P = apollo_combineModels(P, apollo_inputs, functionality)

Have a look at the hybrid examples

Best wishes

Stephane