Derived from the inverted iRAP Decimal Star Rating (DSR), which encodes static geometric risk. A star rating of 1.0 (worst) yields baseline risk 1.0, and 5.0 (best) maps to 0.2:

B(i) = (6 - DSR(i)) / 5

Measures speed deviation and vulnerable road user (VRU) density. Splits speed deviations asymmetrically to separate Congestion Instability (CI) from Overspeed Risk (OR):

SR = 1.0 + 0.4*CI + 0.6*OR

VRU = 1.0 + (Fpeds / Fmax) * 1.5

E(i, t) = SR * VRU (Clamped to [0.5, 5.0])

Applies Network Kernel Density Estimation (NKDE) to historical crash records within a 500m bandwidth. Each crash's impact decays exponentially with age to reflect road modifications over time:

C(i, t) = 1 + ∑ [ws(j) * K(d, h) * e-r * age]

Severity weights: Fatal (5.0), Serious (3.0), Minor (1.0), Property-only (0.5). Decay rate: 30% per year.

Accounts for transient risk factors (weather, lane closures due to construction, active incidents blocking roads):

T(i, t) = max(1.0, Tw * Tcon * Tinc)

Calculate raw multiplicative risk $R(i, t)$, then rescale it to a 0-100 score relative to observed sandbox endpoints ($R_{\text{min}}$ and $R_{\text{max}}$):

DRRS = 100 * (R - Rmin) / (Rmax - Rmin)