105: Swissmetro Normal Mixed MNL Mode ChoiceΒΆ


Mixed logit models are under development. The interface should not be considered stable and may change with future versions of Larch. Use at your own risk.

This example is a mode choice model built using the Swissmetro example dataset. First we create the DB and Model objects:

d = larch.DB.Example('SWISSMETRO')
kernel = larch.Model(d)

We can attach a title to the model. The title does not affect the calculations as all; it is merely used in various output report styles.

kernel.title = "swissmetro example 05 (normal mixture logit)"

The swissmetro dataset, as with all Biogeme data, is only in co format. To add a mixed parameter on time, we give the time parameter twice (once with *1 added to give a distinctive name) and add the plain (mean of the normal) and distributional (std dev of the normal).

kernel.utility.co("TRAIN_TT *1",1,"B_TIME_S")
kernel.utility.co("SM_TT *1",2,"B_TIME_S")
kernel.utility.co("CAR_TT *1",3,"B_TIME_S")

From the kernel MNL model we create a normal mixed model. We set the starting value for the std dev to be nonzero to improve numerical stability:

m = larch.mixed.NormalMixedModel(kernel, ['B_TIME_S'], ndraws=100, seed=0)
v = m.parameter_values()
v[-1] = 0.01

We can estimate the models and check the results match up with those given by Biogeme:

>>> result = m.maximize_loglike()
>>> m.loglike()

The reporting features of mixed logit models have not been developed yet. A placeholder simple report of the parameters is available for now:

>>> print(m)
<larch.mixed.NormalMixedModel> Temporary Report
ASC_TRAIN                       -0.396863
ASC_CAR                          0.140273
B_TIME                          -0.0235885
B_COST                          -0.0128322
Choleski_0                       0.0160842


If you want access to the model in this example without worrying about assembling all the code blocks together on your own, you can load a read-to-estimate copy like this:

m = larch.Model.Example(105)