5 Ways Planners Can Improve Their Models with Streetlytics

Streetlytics harnesses vast amounts of location data to provide accurate information about the movement of people throughout the USA. That data is reviewed, analyzed and then weighted to match against over 1.2 million traffic counts and against current block group level household and employment data to provide a complete understanding of movement and not simply insights from a sample.

Streetlytics provides modelers with several ways to improve traffic forecasting accuracy:

1) Model Validation

Models are applied to base year inputs to calculate base year model estimated origin-destination matrices of movement, and traffic assignments of volumes, speed and congestion.  The model results are then compared and adjusted to match independent observations in a process known as model validation.  Streetlytics provides rich sources of data for model validation:

  • Origin-destination vehicular trip matrices by trip purpose, by season, by day of week, and period of the day
    Destinations of weekday travelers originating at O-Hare airport


  • Highly accurate measurements of traffic volumes at all locations on the road network by season, by day of the week, and hour of the day

    Average annual weekday traffic volumes in Northern Virginia
  • Measured travel speeds on all roadways by day of the week and hour of the day

    Average annual weekday traffic volumes in Northern Virginia

2) External Origin-Destination Matrices

One of the classic weaknesses of any model is its ability to estimate travel that passes through the model region or has one end of the trip outside of the model region.  Models sometimes represent external travel through three sets of origin-destination matrices:

  • Internal to external flows
  • External to internal flows, and
  • External to external flows.

Streetlytics provides external origin-destination matrices.

Origins and destinations of travel entering the Orlando region on US 50

3) Improved Time of Day, Weekend and Seasonal Modeling

For those regions with models that incorporate time of day, weekends and seasons in their models, Streetlytics contains rich observed data:

  • Observed OD patterns by season, period of the day, and day of the week, including weekends
  • Measured traffic volumes on all road segments by season, by hour of the day, day of the week and season.
Hourly traffic volume by Day Type by Season: Buckman Bridge EB in Jacksonville, FL

4) Spatially Accurate Networks and Future-Proof Zone Systems

Streetlytics data is provided on HERE up-to-date spatially accurate road networks and at the US Census block group zone level.  Spatially accurate networks not only provide a much better visual appearance for results, they also are much more accurate both in their distances and travel times, but also in the road attributes such as the number of lanes versus typical historic ‘stick’ networks.  Block group level data provides a sound basis for the future, as the US Census has decided to move future data products associated with the Census Transportation Planning Products (CTPP) program to this level of geography.

Travel volumes on spatially correct all-streets road network – New Orleans, LA

5) Better Forecasts through Incremental Model Application

Traffic forecasting models are designed for the purpose of forecasting traffic flows, ridership, revenues, congestion levels and environmental impacts associated with infrastructure, operation and policy changes. In order to prove their validity, modelers in many locations apply these models to see how well they can model today.  In the process of doing so, many forecasting models are adjusted so much so that their sensitivities to forecasting change are skewed inappropriately.  Streetlytics offers an opportunity to avoid this pitfall by applying forecasting models to estimate change and then applying that change on Streetlytics’ highly detailed observed OD flows and volume observations.  That process, used in many places around the world today, is known as incremental modeling.  Not only does this approach allow observed origin-destination matrices to be developed for forecasting, this approach also enables models to be always up to date.  No longer does a model have a ‘base year’—it is simply a forecasting model applied to current year Streetlytics data.

Streetlytics offers a new opportunity to traffic forecasters to improve their models and their forecasts by harnessing vast amounts of observed calibrated movement data at a very low cost.  In addition to offering Streetlytics directly to agencies and consultants, Citilabs provides full assistance to take advantage of The Streetlytics Opportunity.

For more information or a Streetlytics demo, contact us today.