inst/examples/bayesian_swissmetro/plot_b08_boxcox.R

In rbiogeme: Interface to 'Biogeme'

#!/usr/bin/env Rscript

# b08. Bayesian multinomial logit with a common Box--Cox time transformation.
#
# The expression tree below is complete. boxcox() maps directly to native
# Biogeme's BoxCox expression, including its behavior near lambda = 0.

library(rbiogeme)

# prepare_swissmetro_example() is defined in ../swissmetro/example_utils.R.
# It reads the data and Python executable options and creates the output path.
script_path <- commandArgs(trailingOnly = FALSE)
script_path <- sub("^--file=", "", script_path[startsWith(script_path, "--file=")][[1L]])
source(file.path(dirname(normalizePath(script_path)), "..", "swissmetro", "example_utils.R"))

build_b08_boxcox_model <- function(database) {
  # These starts, bounds, and fixed status values match Bayesian Python b08.
  asc_car <- biogeme_beta("asc_car", start = 0)
  asc_train <- biogeme_beta("asc_train", start = 0)
  asc_sm <- biogeme_beta("asc_sm", start = 0, fixed = TRUE)
  b_time <- biogeme_beta("b_time", start = -1.675, upper = 0)
  b_cost <- biogeme_beta("b_cost", start = -1.079, upper = 0)
  boxcox_parameter <- biogeme_beta(
    "boxcox_parameter",
    start = 1,
    lower = -2,
    upper = 2
  )

  train_time <- boxcox(variable("TRAIN_TT_SCALED"), boxcox_parameter)
  sm_time <- boxcox(variable("SM_TT_SCALED"), boxcox_parameter)
  car_time <- boxcox(variable("CAR_TT_SCALED"), boxcox_parameter)
  utilities <- list(
    `1` = asc_train + b_time * train_time +
      b_cost * variable("TRAIN_COST_SCALED"),
    `2` = asc_sm + b_time * sm_time +
      b_cost * variable("SM_COST_SCALED"),
    `3` = asc_car + b_time * car_time +
      b_cost * variable("CAR_CO_SCALED")
  )
  availability <- list(
    `1` = variable("TRAIN_AV_SP"),
    `2` = variable("SM_AV"),
    `3` = variable("CAR_AV_SP")
  )

  # The Python example uses 10,000 posterior draws and 10,000 warmup draws.
  log_probability <- logit_log_probability(
    utilities,
    availability,
    alternative = variable("CHOICE")
  )
  biogeme_model(
    database = database,
    formula = log_probability,
    control = biogeme_control(
    output_directory = prepared$output,
      model_name = "b08_boxcox",
      bayesian_draws = 10000,
      warmup = 10000,
      generate_html = TRUE,
      generate_yaml = TRUE,
      generate_netcdf = TRUE
    )
  )
}

prepared <- prepare_swissmetro_example(
  commandArgs(trailingOnly = TRUE),
  default_model = "b08_boxcox"
)

# Start a fresh native run; exact old outputs cannot be silently recycled.
unlink(file.path(prepared$output, c(
  "b08_boxcox.yaml",
  "b08_boxcox.nc",
  "b08_boxcox.html",
  "__b08_boxcox.iter"
)), force = TRUE)

database <- swissmetro_data(prepared$data)
model <- build_b08_boxcox_model(database)
fit <- bayesian_estimate(model, model_name = "b08_boxcox", control = model$control)

print(summary(fit))
print(coef(fit))
print(bayesian_stored_variables(fit))
invisible(fit)