TAC programme view — state level first. Ranked by Collective Risk (FSI/km/yr); geography refines, it does not precede value.
Pre-EB shortlist. Statewide top-500 by CR plus every LGA's top-15 (1,326 corridors, ≥1 km). CR basis: whole road — modelled FSI over ALL sections ÷ full road length; cov shows the reliability-ok data share. Speed limits: official zones (May 2026) with connected-vehicle fallback. Operating speeds measured where shown; ΔFSI columns use Elvik (2019) β=0.08, response 0.5 — secondary to CR ranking. Region attribution v1 (LGA→transport region mapping, derived). Not a programme lock-in.
Scenario:
Map — corridors coloured by Collective Risk
Supply curve (CR order)
Walking the CR-ranked list top-down: cumulative ΔFSI/yr saved vs cumulative veh-hours. Direct cost only, no reassignment.
Signals & outliers (measured-speed corridors)
Ranked corridors — CR is the sort
Road (LGA)
km
CR
cov
PR
AADT
limit
v̄
FSI/yr
ΔFSI
cost vh/yr
CR = MODELLED expected FSI/yr per km (NB-GLM full-population layer, temporal holdout obs/exp 0.968; FSI-count outcome, NOT FSI-Equivalent severity-weighted). Whole-road basis: all sections ÷ full length. PR = FSI per 100M VKT (VKT-weighted). Crash points = FSI history 2012–2024 (fatal red, serious orange) — observed history, distinct from modelled CR. ΔFSI = FSI × (1−e^(−β·Δv̄)), β=0.08 (Elvik et al. 2019); Δv̄=0.5×step. Sources: crashstats risk layer, banded AADT, Victorian speed zones May 2026, connected-vehicle speed tiles (Jul 2026). Casualty-savings claims pending EB/CMF gate. crashstats.com.au family.