Skip to contents

Calculate day pattern probabilities

save_daypat_results.Rd

Computes predicted probabilities for day pattern categories (none, mandatory, non-mandatory) using a fitted model and survey data, with and without bias parameter adjustment. Use to compare adjusted and unadjusted predictions, validate rMove platform invariance, and optionally save results for reporting.

Usage

save_daypat_results(adjusted_probs, unadjusted_probs, report_dir)

Arguments

fit_dt

data.table with required columns:

  • diary_platform — survey platform (e.g., 'rmove', 'call', 'online')

  • day_group <character/integer> — day group identifier

  • day_weight — day-level expansion weight

  • estimation_weight — estimation weight for rMove validation

  • diary_call — bias parameter (set to 0 for unadjusted)

  • diary_online — bias parameter (set to 0 for unadjusted) Other predictors as required by model. Rows: one per day. Keys: none required. Modified by reference: no (returns copy).

model

fitted multinomial model (e.g., nnet::multinom). Must support predict(..., type = "probs").

rmove_adjustment

logical(1). If TRUE, checks rMove platform invariance. Default: TRUE.

save_results

logical(1). If TRUE, saves results to CSV. Default: TRUE.

settings

list. Must include:

  • report_dir — directory for saving results (if save_results is TRUE)

Value

data.table. Adjusted probabilities for each day, with columns:

  • prob_made_none — probability of no trips

  • prob_made_mandatory — probability of mandatory trips

  • prob_made_nonmand — probability of non-mandatory trips All original columns from fit_dt included.

Details

  • Copies input data to avoid reference modification.

  • Sets bias parameters to zero for unadjusted predictions.

  • Uses model to predict probabilities for each day pattern category, both adjusted and unadjusted.

  • Compares weighted sums by diary platform and day group, reporting percent differences.

  • Validates that rMove platform predictions are unaffected by adjustment.

  • Optionally saves results to CSV via save_daypat_results() if save_results is TRUE.

  • Returns adjusted probability data.table.

Settings

  • report_dir (direct): Directory for saving results. Default from config/settings.json.

See also

save_daypat_results()

Other day pattern modeling: calc_day_trip_class(), calc_daypat_probabilities()

Examples

## Not run:
# fit_dt <- ... # survey day-level data
# model <- ... # fitted multinomial model
# settings <- list(report_dir = "output")
# probs <- calc_daypat_probabilities(fit_dt, model, settings = settings)
## End(Not run)