Note
Go to the end to download the full example code.
Mixture of logit models
Example of a normal mixture of logit models, using numerical integration.
- author:
Michel Bierlaire, EPFL
- date:
Sun Apr 9 17:33:39 2023
import biogeme.biogeme_logging as blog
import biogeme.biogeme as bio
import biogeme.distributions as dist
from biogeme import models
from biogeme.expressions import (
Beta,
RandomVariable,
log,
Integrate,
)
See the data processing script: Data preparation for Swissmetro.
from swissmetro_data import (
database,
CHOICE,
SM_AV,
CAR_AV_SP,
TRAIN_AV_SP,
TRAIN_TT_SCALED,
TRAIN_COST_SCALED,
SM_TT_SCALED,
SM_COST_SCALED,
CAR_TT_SCALED,
CAR_CO_SCALED,
)
logger = blog.get_screen_logger(level=blog.INFO)
logger.info('Example b05normal_mixture_integral.py')
Example b05normal_mixture_integral.py
Parameters to be estimated.
ASC_CAR = Beta('ASC_CAR', 0, None, None, 0)
ASC_TRAIN = Beta('ASC_TRAIN', 0, None, None, 0)
ASC_SM = Beta('ASC_SM', 0, None, None, 1)
B_COST = Beta('B_COST', 0, None, None, 0)
Define a random parameter, normally distributed, designed to be used for numerical integration.
B_TIME = Beta('B_TIME', 0, None, None, 0)
It is advised not to use 0 as starting value for the following parameter.
B_TIME_S = Beta('B_TIME_S', 1, None, None, 0)
omega = RandomVariable('omega')
density = dist.normalpdf(omega)
B_TIME_RND = B_TIME + B_TIME_S * omega
Definition of the utility functions.
V1 = ASC_TRAIN + B_TIME_RND * TRAIN_TT_SCALED + B_COST * TRAIN_COST_SCALED
V2 = ASC_SM + B_TIME_RND * SM_TT_SCALED + B_COST * SM_COST_SCALED
V3 = ASC_CAR + B_TIME_RND * CAR_TT_SCALED + B_COST * CAR_CO_SCALED
Associate utility functions with the numbering of alternatives.
V = {1: V1, 2: V2, 3: V3}
Associate the availability conditions with the alternatives.
av = {1: TRAIN_AV_SP, 2: SM_AV, 3: CAR_AV_SP}
Conditional on omega, we have a logit model (called the kernel).
condprob = models.logit(V, av, CHOICE)
We integrate over omega using numerical integration
logprob = log(Integrate(condprob * density, 'omega'))
Create the Biogeme object
the_biogeme = bio.BIOGEME(database, logprob)
the_biogeme.modelName = 'b05normal_mixture_integral'
Biogeme parameters read from biogeme.toml.
Estimate the parameters
results = the_biogeme.estimate()
As the model is rather complex, we cancel the calculation of second derivatives. If you want to control the parameters, change the name of the algorithm in the TOML file from "automatic" to "simple_bounds"
*** Initial values of the parameters are obtained from the file __b05normal_mixture_integral.iter
Cannot read file __b05normal_mixture_integral.iter. Statement is ignored.
As the model is rather complex, we cancel the calculation of second derivatives. If you want to control the parameters, change the name of the algorithm in the TOML file from "automatic" to "simple_bounds"
Optimization algorithm: hybrid Newton/BFGS with simple bounds [simple_bounds]
** Optimization: BFGS with trust region for simple bounds
Iter. ASC_CAR ASC_TRAIN B_COST B_TIME B_TIME_S Function Relgrad Radius Rho
0 1 -1 -1 -1 2 6.1e+03 0.16 1 0.25 +
1 0 -0.73 -0.4 -2 3 5.5e+03 0.049 1 0.36 +
2 0.51 -0.95 -1.4 -2.3 2.6 5.4e+03 0.054 1 0.39 +
3 0.51 -0.95 -1.4 -2.3 2.6 5.4e+03 0.054 0.5 -0.15 -
4 0.0062 -0.45 -1.1 -2.8 2.6 5.3e+03 0.03 0.5 0.49 +
5 0.33 -0.089 -1.6 -2.6 2.5 5.3e+03 0.047 0.5 0.14 +
6 0.33 -0.089 -1.6 -2.6 2.5 5.3e+03 0.047 0.25 -0.17 -
7 0.26 -0.34 -1.4 -2.9 2.3 5.2e+03 0.022 0.25 0.66 +
8 0.22 -0.3 -1.2 -2.6 2.2 5.2e+03 0.0086 0.25 0.53 +
9 0.22 -0.3 -1.2 -2.6 2.2 5.2e+03 0.0086 0.12 -3.2 -
10 0.22 -0.3 -1.2 -2.6 2.2 5.2e+03 0.0086 0.062 -0.28 -
11 0.28 -0.36 -1.3 -2.6 2.1 5.2e+03 0.0065 0.062 0.38 +
12 0.22 -0.3 -1.4 -2.6 2.1 5.2e+03 0.0064 0.062 0.54 +
13 0.21 -0.36 -1.3 -2.6 2 5.2e+03 0.0094 0.062 0.49 +
14 0.22 -0.33 -1.3 -2.5 2 5.2e+03 0.0035 0.062 0.89 +
15 0.18 -0.35 -1.3 -2.5 1.9 5.2e+03 0.0055 0.062 0.85 +
16 0.21 -0.36 -1.3 -2.4 1.9 5.2e+03 0.003 0.062 0.65 +
17 0.17 -0.37 -1.3 -2.4 1.8 5.2e+03 0.0021 0.062 0.84 +
18 0.17 -0.39 -1.3 -2.3 1.8 5.2e+03 0.002 0.062 0.58 +
19 0.13 -0.4 -1.3 -2.3 1.7 5.2e+03 0.0039 0.062 0.27 +
20 0.12 -0.41 -1.3 -2.2 1.7 5.2e+03 0.0019 0.062 0.18 +
21 0.12 -0.41 -1.3 -2.2 1.7 5.2e+03 0.0019 0.031 -0.47 -
22 0.13 -0.42 -1.3 -2.3 1.6 5.2e+03 0.0012 0.031 0.73 +
23 0.13 -0.42 -1.3 -2.3 1.6 5.2e+03 0.0012 0.016 -1.3 -
24 0.13 -0.42 -1.3 -2.3 1.6 5.2e+03 0.0012 0.0078 0.066 -
25 0.13 -0.41 -1.3 -2.2 1.7 5.2e+03 0.0011 0.0078 0.36 +
26 0.13 -0.41 -1.3 -2.3 1.7 5.2e+03 0.00016 0.0078 0.81 +
27 0.13 -0.41 -1.3 -2.3 1.7 5.2e+03 0.00016 0.0039 -1.1 -
28 0.13 -0.41 -1.3 -2.3 1.7 5.2e+03 0.00016 0.002 -0.15 -
29 0.14 -0.4 -1.3 -2.3 1.7 5.2e+03 0.0003 0.002 0.36 +
30 0.14 -0.4 -1.3 -2.3 1.7 5.2e+03 7.1e-05 0.002 0.69 +
Results saved in file b05normal_mixture_integral.html
Results saved in file b05normal_mixture_integral.pickle
print(results.short_summary())
Results for model b05normal_mixture_integral
Nbr of parameters: 5
Sample size: 6768
Excluded data: 3960
Final log likelihood: -5214.88
Akaike Information Criterion: 10439.76
Bayesian Information Criterion: 10473.86
pandas_results = results.get_estimated_parameters()
pandas_results
Total running time of the script: (0 minutes 15.375 seconds)