Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Addressing Data Challenges for Bicycle Crash Analysis

212 views

Published on

Eleni Christofa, University of Massachusetts Amherst

Published in: Education
  • Be the first to comment

  • Be the first to like this

Addressing Data Challenges for Bicycle Crash Analysis

  1. 1. Eleni Christofa, PhD Civil and Environmental Engineering University of Massachusetts Amherst TREC Friday Transportation Seminar Series Portland, OR March 10, 2017 Addressing Data Challenges for Bicycle Crash Analysis
  2. 2. 2University of Massachusetts Transportation Center Motivation https://www.flickr.com/photos/infomatique/6210901187
  3. 3. 3University of Massachusetts Transportation Center Motivation Bikesharing Station Demand Data Mobile Bicycle Data Fixed Location Demand Data
  4. 4. 4University of Massachusetts Transportation Center Data Challenges
  5. 5. 5University of Massachusetts Transportation Center Background R = 1, 000, 000A 365V R: crash rate in crashes per million vehicles A: average number of crashes per year V: average volume of vehicles per day, or average annual daily vehicles (AADT)
  6. 6. 6University of Massachusetts Transportation Center Background RIVER STREET MAIN STREET MASSACHUSETTS AVENUE BROADWAY HAMPSHIRE STREET JOHN F KENNEDY STREET VASSAR STREET ALEWIFE BROOK PARKWAY CAMBRIDGE STREET MOUNT AUBURN STREET BRATTLE STREET GARDEN STREET WESTERN AVENUE BROOKLINE STREET HURON AVENUE QUINCY STREET 0 250 500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,500 2,750 3,000 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 22,500 AADT AADB
  7. 7. 7University of Massachusetts Transportation Center Background 0 25 50 75 100 125 150 175 200 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 2,200 2,400 2,600 2,800 3,000 AADB BicycleCrashFrequency 0 25 50 75 100 125 150 175 200 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 18,000 20,000 22,000 24,000 26,000 28,000 30,000 AADT BicycleCrashFrequency
  8. 8. 8University of Massachusetts Transportation Center Objective §  To assess bicycle crash risk accounting for: §  double exposure to both cars and other bicycles §  data challenges §  seasonality in bicycle demands §  lack of continuous counts in multiple locations
  9. 9. 9University of Massachusetts Transportation Center “Double Exposure” Crash Rate Rdual: crash rate in crashes per million vehicles A: average number of crashes per year Vauto: average volume of automobile traffic per day, or average annual daily vehicles (AADT) Vbike: average volume of bicycle traffic per day, or average annual daily bicycles (AADB) Rdual = ✓ 1, 000, 000 365 ◆2 · A VautoVbike
  10. 10. 10University of Massachusetts Transportation Center Research Approach: Framework AADB Estimation Corridor Assignment Crash Analysis
  11. 11. 11University of Massachusetts Transportation Center Count data Automobile: AADT from 2012 (MassDOT) Bicycle: Manual peak hour count data at 28 locations City of Cambridge: 2-hr AM and PM peaks (3 days in Sept. 2012) Boston MPO: 1-4 hrs (2009-2014) Continuous count data at 2 locations Broadway Avenue (November 2013 – June 2014) Hampshire Street & Cardinal Medeiros Avenue (July 2015 – 2016) Source: City of Cambridge, MA; Boston MPO
  12. 12. 12University of Massachusetts Transportation Center Bicycle Crash Data Location: Cambridge, MA Time Interval: 2011-2014 Number of crashes: 622 bicycle-vehicle crashes Source: UMass Safety Data Warehouse Source: https://velosurance.com
  13. 13. 13University of Massachusetts Transportation Center 1. Annual Average Daily Volume Estimation: A Sinusoidal Bicycle Demand Model
  14. 14. 14University of Massachusetts Transportation Center Bicycle Demand Estimation: Research Approach 1.  Bicycle counts and bike-share data 2.  Data analysis 3.  Model calibration 4.  Model validation
  15. 15. 15University of Massachusetts Transportation Center 1. Bicycle Counts City Count Locations Ottawa, ON 12 Cambridge, MA 1 Arlington, VA 21 Portland, OR 6 Vancouver, BC 4 Seattle, WA 3
  16. 16. 16University of Massachusetts Transportation Center 1. Bike-Share Data City Bike-Share Name Available Data Boston, MA Hubway Bike-Share 2011-2013 Washington D.C. Capital Bike-Share 2013-2015 New York City, NY Citi Bike-Share 2010-2015 Saint Paul, MN Nice Ride Bike-Share 2010-2015
  17. 17. 17University of Massachusetts Transportation Center 2. Data Analysis 0 500 1,000 1,500 2,000 Jun-11 Sep-11 D ec-11 M ar-12 Jun-12 Sep-12 D ec-12 M ar-13 Jun-13 Sep-13 D ec-13 M ar-14 Jun-14 Sep-14 D ec-14 M ar-15 MonthlyADB 2011 2012 2013 2014 Estimated ADB Laurier Ave. and Metcalfe St., Ottawa, ON
  18. 18. 18University of Massachusetts Transportation Center 3. Model Calibration Portland, OR
  19. 19. 19University of Massachusetts Transportation Center 3. Model Calibration Seattle, WA
  20. 20. 20University of Massachusetts Transportation Center 3. Model Calibration: Sinusoidal Function Month, 𝑡 0.0 0.2 0.4 0.6 0.8 1.0 1 2 3 4 5 6 7 8 9 10 11 12 𝑊𝑎𝑣𝑒𝑙𝑒𝑛𝑔𝑡ℎ, 𝜔=​π⁄6  Amplitude, A Centerline Average, AADB 𝑊𝑎𝑣𝑒 𝐶𝑟𝑒𝑠𝑡, ADBMax 𝑊𝑎𝑣𝑒 𝑇𝑟𝑜𝑢𝑔ℎ, ADBMin NormalizedMonthlyADB Phase Shift, φ
  21. 21. 21University of Massachusetts Transportation Center 3. Model Calibration: Sinusoidal Function MADBt = AADB + A · sin(! · (t )) MADBt: Monthly Average Daily Bicycle Count for month t, [bicycles/day] AADB: sinusoidal centerline [bicycles/day] t: time value [months] A: amplitude of MADB sinusoid, or the average seasonal change [bicycles/day] ω =2πf: wavelength of MADB sinusoid [months] f=1/12: frequency of MADB sinusoid φ: phase of sinusoid
  22. 22. 22University of Massachusetts Transportation Center 3. Model Calibration: Sinusoidal Function A = MADBMax MADBMin 2 AADB = P365 i=1 ADBt 365 AADB = MADBMax + MADBMin 2 or
  23. 23. 23University of Massachusetts Transportation Center 3. Model Calibration: Seasonal Change α
  24. 24. 24University of Massachusetts Transportation Center 3. Model Calibration: Seasonal Change a = MADBMax MADBMin MADBMax + MADBMin City Count Locations Temp. Difference (oF) (High-Low) α Ottawa, ON 11 56.6 (70.2-13.6) 0.96 Cambridge, MA 1 44.4 (73.4-29.0) 0.58 Arlington, VA 20 43.8 (79.8-36.0) 0.61 Portland, OR 6 29.1 (69.5-40.4) 0.45 Vancouver, BC 4 25.9 (64.4-38.5) 0.62 Seattle, WA 3 24.1 (66.1-42.0) 0.55
  25. 25. 25University of Massachusetts Transportation Center 3. Model Calibration: Seasonal Change AADB = MADBt ↵ · sin(⇡ 6 (t )) + 1 MADBt = AADB h ↵ · sin ⇣⇡ 6 (t ) ⌘ + 1 i
  26. 26. 26University of Massachusetts Transportation Center Bike-share demand data alpha = 0.52 alpha = 0.37 alpha = 0.39 alpha = 0.45 alpha = 0.44 2011 2012 2013 2014 2015 0 2,500 5,000 7,500 10,000 12,500 15,000 Jun-10 Sep-10 D ec-10 M ar-11 Jun-11 Sep-11 D ec-11 M ar-12 Jun-12 Sep-12 D ec-12 M ar-13 Jun-13 Sep-13 D ec-13 M ar-14 Jun-14 Sep-14 D ec-14 M ar-15 Jun-15 Sep-15 D ec-15 M ar-16 MonthlyADB 2010 2011 2012 2013 2014 2015 AADB Estimated ADB Capital Bike-share Data, Washington D.C.
  27. 27. 27University of Massachusetts Transportation Center Bike-share demand data Bike-share System α Hubway (Boston, MA) 0.69 Citi Bike (NY, NY) 0.63 Capital Bike (Washington, DC) 0.49 Nice Ride (Saint Paul, MN) 0.78
  28. 28. 28University of Massachusetts Transportation Center 4. Model Validation Ottawa, Canada, α = 0.99 Month MADB
  29. 29. 29University of Massachusetts Transportation Center 4. Model Validation Arlington, VA, α = 0.57 Month MADB
  30. 30. 30University of Massachusetts Transportation Center Estimation Accuracy •  AADB errors varied between 8.64% (August) and 28.11% (January) for all cities •  MADB errors varied between 5.36% (July) and 41.54% (January) for all cities
  31. 31. 31University of Massachusetts Transportation Center 2. Corridor Assignment
  32. 32. 32University of Massachusetts Transportation Center Map of Bicycle-Vehicle Collisions, Cambridge, MA 9 8 7 6 5 4 3 2 1 0 17 16 15 14 13 12 11 10 0 0.5 1 1.5 20.25 Miles Cambridge Bicycle-Vehicle Collision Map, 2011-2014 Legend Study Intersection Bicycle Collision Bicycle/Pedestrian priority Cycle track Bike lane; Hybrid Marked shared lane On-Road/No Lane 0 - Inman Square 1 - Massachusetts Ave & Vassar St 2 - Broadway & Hampshire St 3 - Massachusetts Ave & Memorial Drive 4 - Lafayette Square 5 - JFK St & Memorial Drive 6 - Porter Square 7 - Brattle St at Sparks St & Craigie St 8 - Western Ave & Memorial Drive 9 - Massachusetts Ave & Linear Park 10 - Brookline St & Granite St 11 - Quincy Square 12 - Brattle St & Mason St 13 - Fresh Pond Parkway & Concord Ave 14 - Arsenal Square 15 - Huron Ave & Fayerweather St 16 - River St & Putnam St 17 - Hampshire St & Cardinal Mederios Ave
  33. 33. 33University of Massachusetts Transportation Center Bicycle Corridors for Cambridge, MA ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! !!! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! BROOKLINE BELMONT ARLINGTON SOMERVILLE WATERTOWN BOSTON MEDFORD BROADWAY CAMBRIDGE STREET BRATTLE STREET VASSAR STREET M ASSACHUSETTS AVENUE RIVER STREET GARDEN STREET HAM PSHIRE STREET BROOKLINE STREET MAIN STREET ALEWIFE BROOK PARKWAY HURON AVENUE MOUNT AUBURN STREET M ASSACHUSETTS AVENUE! Bicycle Crash Location Assumed Bicycle Corridor Street without Bicycle Facilities Street with Bicycle Facilities Bicycle Count Location
  34. 34. 34University of Massachusetts Transportation Center 3. Crash Risk Analysis
  35. 35. 35University of Massachusetts Transportation Center “Double exposure” vs Conventional Crash Rates R = 1, 000, 000A 365V Rdual = ✓ 1, 000, 000 365 ◆2 · A VautoVbike
  36. 36. 36University of Massachusetts Transportation Center “Double Exposure” Crash Rate & Crash Frequency Combined Rate 2 4 6 8 Crash Frequency 1 2 3 Crash frequencies along corridors Crash rates along corridors
  37. 37. 37University of Massachusetts Transportation Center Conclusions: Bicycle Crash Rate Innovative components: 1.  accounts for bicycle exposure to both automobiles and bicycles 2.  addresses data challenges through: •  a seasonal bicycle demand model •  corridor-based analysis Limitations: •  Uncertainty in the impact of automobiles vs bicycles on bicycle crash risk
  38. 38. 38University of Massachusetts Transportation Center Conclusions: Seasonal Demand Model Advantages: •  Can estimate MADB and AADB using only two short-term counts Considerations: •  Lack of seasonality •  Cyclist type •  Low counts
  39. 39. 39University of Massachusetts Transportation Center Other Ongoing Projects
  40. 40. 40University of Massachusetts Transportation Center
  41. 41. 41University of Massachusetts Transportation Center Acknowledgments Nick Fournier and Mike Knodler Funding: •  SAFER-SIM UTC & New England UTC •  Eisenhower Graduate Transportation Fellowship Data: •  UMass Safe Traffic Safety Research Program •  Portland, OR; Arlington, VA, Seattle, WA; Ottawa, ON; Vancouver, BC; Cambridge, MA
  42. 42. 42University of Massachusetts Transportation Center References 1.  Fournier, N., Christofa, E., and Knodler, M.A. 2017. A Mixed Method Investigation of Bicycle Exposure in Crash Rates, Accident Analysis & Prevention. [in press] 2.  Fournier, N., Christofa, E., and Knodler, M. A. 2017. A sinusoidal model for seasonal bicycle demand estimation. Transportation Research Part D: Transport and Environment, 50, 154-169.
  43. 43. 43University of Massachusetts Transportation Center Questions? Eleni Christofa christofa@ecs.umass.edu

×