Developing a Bicyclist Route Choice Model Using GPS Data

Developing a bicyclist route choice model

Urban Studies and PlanningJoe Broach, PhD StudentJohn Gliebe, ProfessorJennifer Dill, Professor

Random Utility Model

AlternativeRoutes

Attributes Weights

Choice

Stated Preference

Tilahun et al. (2007)

Revealed Preference

GPS TrackBike Network

no facilityAttributes

signed bike routeon-street bike lanebike boulevardoff-street path

Cyclists 154Women 44%Year-round 96%

Trips 1464Commute 30%

Miles 5169Portland 83%

GPS Data used for model (2007)

Alternative generation

GPS TrackBike Network

no facilityAttributes

signed bike routeon-street bike lanebike boulevardoff-street path

Calibrated Labeling Method

Optimize key attributes

Subject to…

Maximize bike lane use, subject to distance constraint

Sample Alternative (1)observed

alternative

Sample Alternative (2)

Relaxing distance constraint results in 2nd

unique alternative.

2nd alternative

Sample Alternative (3)

Relaxing distance constraint further results in 3rd unique alternative.

3rd alternative

Attributes

Facility Intersection Physical

Bike FacilityFunctional ClassTraffic VolumeOne-way

TurnsSignal/StopCross Volume

DistanceSlope

Bridge path

Bridge bike lane

Bike path

Bike boulevard

Bike lane

Minor arterial

Major arterial

Highway

Equivalent Distances by Facility (90 percent random param. intervals)

Questions?

jbroach@pdx.edu