Mixtures of logit with Monte-Carlo 10_000 MLHS draws

Estimation of a mixtures of logit models where the integral is approximated using MonteCarlo integration with MLHS draws.

Michel Bierlaire, EPFL Tue Apr 29 2025, 12:17:03

from b07estimation_specification import get_biogeme
from IPython.core.display_functions import display

import biogeme.biogeme_logging as blog
from biogeme.expressions import Draws
from biogeme.results_processing import (
    EstimationResults,
    get_pandas_estimated_parameters,
)

logger = blog.get_screen_logger(level=blog.INFO)
logger.info('Example b07estimation_monte_carlo_mlhs.py')

Example b07estimation_monte_carlo_mlhs.py

R = 10_000

the_draws = Draws('b_time_rnd', 'NORMAL_MLHS')
the_biogeme = get_biogeme(the_draws=the_draws, number_of_draws=R)
the_biogeme.model_name = 'b07estimation_monte_carlo_mlhs'
results_file = f'saved_results/{the_biogeme.model_name}.yaml'

Biogeme parameters read from biogeme.toml.

try:
    results = EstimationResults.from_yaml_file(filename=results_file)
except FileNotFoundError:
    results = the_biogeme.estimate()

print(results.short_summary())

Results for model b07estimation_monte_carlo_mlhs
Nbr of parameters:              5
Sample size:                    10719
Excluded data:                  9
Final log likelihood:           -8571.055
Akaike Information Criterion:   17152.11
Bayesian Information Criterion: 17188.51

Get the results in a pandas table

pandas_results = get_pandas_estimated_parameters(
    estimation_results=results,
)
display(pandas_results)

        Name     Value  Robust std err.  Robust t-stat.  Robust p-value
0  asc_train -0.467310         0.047975       -9.740668    0.000000e+00
1     b_time -1.862710         0.075854      -24.556570    0.000000e+00
2   b_time_s  1.206771         0.087492       13.793007    0.000000e+00
3     b_cost -0.845475         0.057652      -14.665121    0.000000e+00
4    asc_car  0.180120         0.035150        5.124271    2.986908e-07