Fitting the Biogeme 01logit model with DiscreteChoiceModels.jl, we do not get the same coefficients as in Biogeme. The initial log-likelihood is the same, but the final log-likelihood is not, the pseudo-R2 is lower, etc. When setting starting values to the final values from Biogeme, the initial log-likelihood is quite smaller.
Fitting the Biogeme 01logit model with DiscreteChoiceModels.jl, we do not get the same coefficients as in Biogeme. The initial log-likelihood is the same, but the final log-likelihood is not, the pseudo-R2 is lower, etc. When setting starting values to the final values from Biogeme, the initial log-likelihood is quite smaller.