Mapping Bike Traffic Data: Revealing Infrastructure Usage Patterns

A common saying in the transportation world is “if you build it, they will come”. Measuring the impact of new bike infrastructure on cycling traffic and behaviour is top of mind for many planners and advocacy groups.

It’s now possible to say with greater accuracy how cycling behaviour is impacted by cycling infrastructure. Using count data to classify cyclist behaviour, Eco-Counter’s Fraser McLaughlin mapped out bike traffic data in Arlington, Virginia per average behaviour patterns.

Behaviour is categorized as either commuter, recreational or mixed by using a type of statistical analysis called cluster analysis. If you missed Fraser’s presentation on cluster analysis research at Bike Hack Night in Washington D.C. (pictured), read on to learn more.

In Arlington, Virginia, automatic counters were first installed in 2009 at two locations to measure bike and pedestrian traffic 24/7. Today, there are more than 30 counters and the data can be viewed on their website. Daily data from 20 permanent counters gathered between April and October in 2016 were analyzed using the following procedure:

  1. An AMI (morning/afternoon index) score for each site measures the intensity of rush hour bike traffic. A low score means traffic is consistent on an hourly basis and a high score means traffic peaks at rush hour when compared to mid-day traffic.
    The difference in bike traffic between rush hour and mid-day
  2. A WWI (weekend/ weekday index) score at each site measures the degree to which bike traffic is higher on the weekend verses weekdays. A high score means a larger amount of cyclist traffic during the weekend when compared to weekday traffic.
    Week day and weekend bike traffic
  3. Using cluster analysis, sites with similar AMI and WWI indexes are grouped together and classified as either predominantly recreational, commuter or mixed traffic sites.

Cluster analysis on bike traffic data

Once sites are classified by how they are used on average, they can be mapped out geographically. The strongest commuter traffic patterns were found at bridges connecting Virginia and D.C. Going in to the analysis, higher commuter traffic was expected at painted on-road bike facilities, yet they were found to be predominantly recreational in use.

Did high traffic sites tend to belong to the same class? The answer is no, locations with over 2000 bikes per day in high season were found in all three usage classifications.

Classified bike traffic in Arlington

How can this information be used? There are several ways in which these indexes and cluster analysis can be applied:

  • When interpolating data, or filling in missing data using what is available: Choosing counters to act as references in the same usage class as the site where data is missing improves the accuracy of the data construction.
  • When extrapolating data, or estimating long term data at a temporary count location: Accuracy is increased when the temporary counter and nearby permanent counters used in the extrapolation are in the same class.
  • When performing before and after comparisons of infrastructure: Change in cyclist behaviour can be more accurately measured before and after infrastructure is added using cluster analysis. This process relies on accurately extrapolating short duration counts taken at the site of the new infrastructure.

Watch for updates on Eco-Counter’s continuing research and development in cluster analysis. Contact our client services team for any general inquiries on our traffic monitoring services and expertise.

Photo credit: M.V. Jantzen

 

Author: Eco-Counter


If you like it, share it!
Facebooktwittergoogle_pluslinkedinmail

Leave a Reply

Your email address will not be published. Required fields are marked *