The initial likelihood is the likelihood for the initial values of the parameters.
If you run several times Biogeme in the same directory, it saves previous estimates, and starts again from there, in order to save time.
- calculate the “null log likelihood”:
where av is the dictionary with the availability conditions.
It only depends on the data and the choice set.
- Disable the automatic saving of the parameters:
biogeme.saveIterations = False
See documentation here:
Make sure also to remove any file that looks like __mymodel.iter in the working directory.