Predictive models of road reliability for traffic sensor configuration and ...

Heterogeneous data sources

(e.g., Logged GPS, road sensors,

Road - segment
case library

706

events, highway incidents, calendar, clock, etc.)

Computed

relationships & properties

Machine learning

rPredictive models

Model 1: Assigns structure- sensitive
mean road speeds to unsensed roads

Model 2: Assigns structure- sensitive
road speed variances to unsensed
roads

Route request

• Directions A -3B

Preferences (Minimal time, distance, risk preferences)

• Real time vs. future (date, time)