Big Data and VMT Laura Schewel laura schewelstreetlightdata

Big Data and VMT Laura Schewel (laura. schewel@streetlightdata. com) Street. Light Data TRB App Con 2017 @streetlightdata

Agenda I. II. VMT Matters & Big Data Matters Examples of how Big Data Advances VMT-oriented policy + planning I. Do traditional VMT measurements better and faster II. Measure VMT in smaller places than before III. Measure VMT in broader places than before IV. Measure VMT more often than before III. Q&A Street. Light In. Sight Web App --Proprietary and Confidential-- 2

Introductions

VMT Matters VMT Trends since 1970 --Proprietary and Confidential-- 4

Data from Mobile Devices Can Show Transportation Experts When, Where, and How People Move (A Subset of Mobile Data from Sept 2016 in Fremont, CA) Key Benefits: • Accurate: Empirically Measures Travel Behavior • Precise: Spatial and Temporal Precision • Comprehensive: Large Sample of Complete Trips and Activities Location-Based Services Data Location Circle radii vary: they accurately reflect the spatial precision of each unique data point Navigation-GPS Data Location Circle enlarged for visibility Note: This image shows a filtered subset of data to improve visibility --Proprietary and Confidential--

But Without the Right Processing Techniques, Big Data is Not Useful Information – Instead, It is A Mess The Information Explosion We help you close the analysis gap by turning Big Data into easy-to-use Metrics. Every month, we shrink ~7. 5 terabytes of location data into travel pattern Metrics designed for transportation. Source for chart: Kay, David and Mark van Harmelen, Delivering benefits from the data deluge , Dec 2012, jisc. ac. uk --Proprietary and Confidential-- This allows our clients to benefit from Big Data – and bypass the labor-intensive processing required to use it effectively. 6

Our Data Processing Engine Turns Messy Data into Useful Transportation Metrics– Without the Hassle Input: Big Data Processing: Route Science® Technology Output via Web App: Street. Light In. Sight Metrics Anonymous and accurate Location Data Road network, land use, parcel, census and more Contextual Data --Proprietary and Confidential-- 7

The Street. Light In. Sight Platform Provides Access to Useful Big Data Analytics for A Range of Transportation Projects and Dem ment age Man sit Tran g/ in n Plan ibility s s Acce ct Proje ion at Evalu ce rman o f r e s P sure Mea ht Freig ng eli Mod --Proprietary and Confidential-- ial Spec orts, s (P Zone s, etc. ) e Bridg y/ ualit Q r i A GHG s ate Estim dor Corri s ie Stud t. spor Tran and Dem ing el Mod t spor Tran ty Equi nal/ Inter al rn Exte es i Stud 8

Using Big Data to Measure VMT

Methodology VMT = Volume * Distance Segment Length T AADT 0 22 1500 A ADT 0 25 T AD A 0 * Share of Trips 1800 AD 0 A Average Trip VMT Miles --Proprietary and Confidential-- 10

Methodology VMT = Volume * Distance Segment Length T AADT 0 22 1500 A ADT 0 25 T AD A 0 * Share of Trips 1800 AD 0 A Average Trip VMT Miles Avg. Trip Distance + Distrubtion are Long Time Street. Light In. Sight Metrics --Proprietary and Confidential-- 11

Methodology VMT = Volume * Distance Segment Length T AADT 0 22 1500 A ADT 0 25 T AD A 0 * Share of Trips 1800 AD 0 A Average Trip VMT Miles AADT is a NEW Metric Coming in June/July. You Get a Preview Today. --Proprietary and Confidential-- 12

Use Case 1 – Measure VMT Faster, More Accurately, More Often, with More Granularity VS VS …. . 3 years ago on a “similar” road Get Volume + Length in a Few Clicks, Anywhere --Proprietary and Confidential-- 13

Use Case 2 – Articulate Variation in VMT for Traffic Impact Estimation (WIP w/ Fehr & Peers) Experiment – Calculate VMT by Land Use for All TAZs in LA SB 743 requires estimation of VMT by land use, especially for residential, office, and retail purposes. BUT – different cities and neighborhoods within cities may vary significantly. Plus, mixed use estimation is not covered. Before Big Data, this was “too big” to measure. --Proprietary and Confidential-- 14

Use Case 3 – (Unexpected “Small” Use Case) “QA” 1 ST CUT - STREETLIGHT AADT ESTIMATE VS. XDOT HIGH FIDELITY LOOP COUNTERS R 2 = 0. 92, Average Absolute Error = 22% 100 000 XDOT AADT (LOG) Log scale on graph Green Line = Unity 10 000 Technique holds for roads from 400 AADT up to 100, 000 AADT+ 1 000 100 000 St. L Estimated AADT (LOG) --Proprietary and Confidential-- 15

Use Case 3 - Anomaly Investigation: Street. Light Technique “QAed” MNDot Data • Counter 46: “CSAH 1 -. 5 MI S OF CSAH 24, W OF WEST CONCORD” XDOT reported AADT of 421 Street Light Estimated >2, 000 Location specified by XDOT Lat/Long (red circle) IS 0. 5 mi South of CSAH 24, but Google shows it on a road numbered 56 IN West Concord not “West of” West Concord What could cause the discrepancy? 16

Use Case 3: QA Continued Going back to the old county maps of Dodge County found: Actual location of the counter Location XDOT said the counter was at The location XDOT said the counter was at is in a town and has MUCH more traffic than rural road where the counter actually is.

Use Case 4 – Before and After (and After) VMT Trends since 1970 --Proprietary and Confidential-- 18

Q&A Want to learn more? Email info@streetlightdata. com or check out our blog and whitepapers at www. streetlightdata. com …. or come visit us in our BOOTH in the TRB App Con Exhibitor hall

Appendix

Our Product Offering is Designed for Transportation Street. Light In. Sight Core Metrics Premium Metrics Sites and Travel Projects For Richer Insights Origin-Destination Analysis Describes trips between Origin and Destination Zones Origin-Destination to Pre-set Geography Describes trips between Zones and either ZIPs, TAZs, or Census Block Groups Trip Attributes Trip time, trip length, trip speed, trip circuity, heavy duty vs. medium duty trucks Traveler Attributes (Combine with O-D for Routing) Describes trips that pass through Middle Filters, or links, between Origin and Destination Zones Simple trip purpose, household income, race, education level of head of household, and family status Zone Activity Analysis Commercial Tours Origin-Destination with Middle Filter Describes trips that originate in, have destinations in, or pass through each Zone analyzed Home and Work Footprints Describes the likely home and work locations of visitors to specific locations, including tourists --Proprietary and Confidential-- (Not Yet in App) The relative volume of tours (within 24 hours) and dwell times at each intermediary stop for tours that begin or end in specified Zones Travel Project Options In Development Metrics to Your Specifications Custom Day Types Define Weekends, Weekdays Custom Day Parts 6 day parts; customize with 1+ hour bins Data Period Choose months to analyze from 2014 to 1 -2 months ago Vehicle Type Commercial vs. Personal Calibration Feature [BETA] Automatically scale to estimated counts by inputting your local ADT, AADT, or similar data Bike Metrics [Special Delivery Timing] Analyze origins and destinations of bike trips Pedestrian Metrics [Special Delivery Timing] Analyze origins and destinations of bicycle trips 2016 AADT Estimate Other extensions of AADT coming soon after 21

Conventional Data Collection and Modeling Tools No Longer Meet All The Needs of Today’s Transportation Professionals Household & Intercept Surveys Aerial Photos & Videos Bluetooth & Other Sensors Assumption. Based Modeled Data --Proprietary and Confidential-- • Expensive • Time-Consuming • Conducted Rarely and/or on a Limited Basis • Small Sample Sizes and/or Incomplete Information • Cumbersome Data Integration The Data Do Not Fully Describe Current Travel Patterns The Data Do Not Persistently Measure Changes Over Time The Data Are Rarely Diagnostic Or Predictive 22

Assumptions– We Made Conservative Estimates Full Budget for Named Projects ~$5 M Annual Street. Light In. Sight Subscription Fee vs. Annual Data Spending for a Mid-Size MPO Street. Light In. Sight Regional Subscription Fixed Annual Fee Multi-Domain Licenses; Premium Metrics for Population of 3. 2 M Household Survey Cost Savings Added Value Data Collection Speed & Unlimited Analyses Survey cost $1. 5 M. Assume 5 yr amortization, 40% displaced. (Last survey cost $1. 5 M. Assume data was 2/3 of costs, and costs were amortized over 5 years. ) Transportation Studies for Modeling Full budget $1. 3 M/yr. Assume 25% displaced. (MPO budgeted $1. 3 M. Assume 1/4 can be displaced. ) Understanding Regional Trucking Flows Assume half displaceable. (MPO budgeted ~$200 k for GPS data biennially. ) TDM for Employer Support Assume only 10% for data from $1 M (MPO budgeted $1 M. Assume 10% is for data. ) Assume 20% for equity-focused data collection. Regional Mobility Hub Implementation (MPO budgeted $413 k. Assume 20% for equity-focused data collection. ) Assume all displaced. Special Studies: Commutes, Corridors, etc. (Data budget estimated based on prior special studies. ) $0 --Proprietary and Confidential-- $200 000 $400 000 $600 000 $800 000 $1 200 000 $1 400 000 $1 600 000 $1 800 000 $2 000 Source: Final FY 2016 SANDAG Program Budget (including the overall work program). Published July 2015. www. sandag. org/uploads/publicationid_1957_19285. pdf 23