Skip to contents

Validate adjusted OD flows against a benchmark overall

Source: R/validate_flows.R
validate_flow_overall.Rd

Compares bias-adjusted MPD flows to benchmark (e.g., census) OD flows. Uses adjusted flows as estimates (x) and benchmark as targets (y), returning summary fit metrics that capture both concordance (for example, correlation) and aggregate error (for example, RMSE, MAE, MAPE). For OD-level residual auditing, use validate_flow_residuals() or the lower-level validate_flow_pairs() table.

Retained for backwards compatibility. New code should prefer validate_flow_overall().

Usage

validate_flow_overall(
  adj_df,
  benchmark_od_df,
  flow_col_adj = "flow_adj",
  flow_col_bench = "flow",
  drop_zeros = TRUE,
  na_rm = TRUE,
  by_source = FALSE,
  return_joined = TRUE,
  method_name = NA_character_
)

validate_flow_benchmark(...)

Arguments

adj_df

Data frame with at least: origin, destination, and a column of adjusted flows (default name "flow_adj"). If present, an mpd_source column is carried through.

benchmark_od_df

Data frame with at least: origin, destination, and a column of benchmark flows (default name "flow").

flow_col_adj

Name of adjusted flow column in adj_df. Default "flow_adj".

flow_col_bench

Name of benchmark flow column in benchmark_od_df. Default "flow".

drop_zeros

Logical, drop rows where either x or y == 0 before metrics. Default TRUE.

na_rm

Logical, remove non-finite rows before metrics. Default TRUE.

by_source

Logical, if TRUE and mpd_source exists in both inputs (or in adj_df), compute metrics per mpd_source as well as overall. Default FALSE.

return_joined

Logical, return the joined row-level data in the result list. Default TRUE.

method_name

Optional label for the adjustment method (e.g. "adjust_inverse_penetration", "adjust_selection_rate"). Stored in the output for comparison workflows.

...

Arguments passed to validate_flow_overall().

Value

A list with:

  • method (if provided)

  • n, sum_adj, sum_bench

  • pearson_r, spearman_rho

  • rmse, mae, mape

  • ols_intercept, ols_slope, r_squared (from lm(y ~ x))

  • (optional) by_source: a tibble of per-source metrics when by_source = TRUE

  • (optional) data: the joined tibble used for the calculations