Skip to contents

Four figures that together tell the fused-OD story: how the sources combine, what that changes structurally, what it changes in amplitude, and what it looks like geographically.

Usage

plot_mobility_fused(
  PATHS,
  suffix = "_fused_raked_tt",
  out_dir = NULL,
  verbose = TRUE
)

Arguments

PATHS

List of paths from get_paths.

suffix

Which est_mobility run to read for the fused quantities. Default "_fused_raked_tt".

out_dir

Directory for the PNGs. Default PATHS$DOCS_FIGURES.

verbose

Print each file written.

Value

Invisibly, a character vector of the files written.

Figures

fused_od_1_sources.png

Small multiples of the four row-normalised source matrices and the fused result, shared sequential scale. Shows what each source contributes.

fused_od_2_landneighbour_share.png

Per-country share of outflow reaching a land neighbour, air vs fused. The single scalar that captures what the method fixes.

fused_od_3_tau_prior.png

Daily departure probability, air fit vs overland prior (the rake target) with 95% intervals, log scale. Read from the unsuffixed prior files, so suffix does not affect it.

fused_od_4_corridors.png

Dominant destination per origin drawn on the map, air vs fused.

Design

Colour follows the data's job: a single-hue sequential ramp for magnitude (heatmaps), and two categorical hues for the air-vs-fused contrast. The categorical pair was validated for colour-vision deficiency (worst-pair protan dE 24.7, normal-vision dE 33.6, both well clear of the 8 / 15 floors), and every two-series panel also carries position or direct labels so identity never rests on colour alone. Countries are ordered by latitude in the heatmaps, matching plot_mobility, so the diagonal band is geographic adjacency.