population_impact_summary(), split by equity band – e.g. “of the people whose journey improved, what share of each deprivation band actually benefited?”. This is what distinguishes a candidate that helps everyone roughly equally from one that wins on the region-wide average while mostly helping people who were already well served (or, just as importantly, one where the most disadvantaged areas are the ones getting worse journeys) – neither is visible from population_impact_summary()’s single region-wide total.
The rate-normalised columns (proportion_of_band_improved/ proportion_of_band_worsened – share of THIS band’s own population, not share of the region-wide improved-total) are the headline: a band with more people can look like it “benefits most” on a raw headcount alone even if only a small fraction of that band actually improved.
Parameters
Name
Type
Description
Default
matrix
Passed straight through to the same frame-building logic as population_impact_summary().
None
demand
Passed straight through to the same frame-building logic as population_impact_summary().
None
meaningful_change_threshold
Passed straight through to the same frame-building logic as population_impact_summary().
None
config_a
Passed straight through to the same frame-building logic as population_impact_summary().
None
config_b
Passed straight through to the same frame-building logic as population_impact_summary().
None
Returns
Name
Type
Description
pandas.DataFrame
Indexed by equity band (raw value, e.g. IMD decile 1-10), ordered from most- to least-disadvantaged per add_equity_data(disadvantaged_end=...) (ties within a tertile keep ascending band order; bands are listed individually, not bucketed into thirds). Columns: regions_improved/ regions_worsened/regions_unchanged, demand_improved/ demand_worsened/demand_unchanged (NaN if no demand data is registered), band_total_demand, proportion_of_band_improved/ proportion_of_band_worsened.
Raises
Name
Type
Description
ValueError
If no equity data is registered on the problem – call add_equity_data() first – or any of the errors population_impact_summary() raises (mismatched demand locations between set_a/set_b, or an unknown secondary demand scenario).