Document Type


Publication Date



Bicycle traffic flow -- Measurement -- Quality control, Bicycling -- Measurement, Pedestrians, Transportation -- Planning, Traffic flow -- Measurement -- Quality control


Cycling and walking are sustainable modes of transportation which improve community livability, but these modes have not been studied with the quantitative rigor applied to motor vehicle travel. This work aims to change that by improving bicycle and pedestrian traffic monitoring data quality, including how erroneous data can most accurately and efficiently be identified through automated processes. The research approach analyzes continuous bicycle and pedestrian count data stored in BikePed Portal, a data archive of bicycle and pedestrian count data. A primary goal of the project is to explore tests that could help to identify aberrant and/or erroneous data. A key method deployed in pursuit of this exploration was to comb through a selection of count data (generally – continuous counter locations from 2015 to 2016 with at least 30 days of counts) to identify expected count ranges and patterns, overall and broken down by rough expected volume levels, along with counts on the fringe or tail end of expected ranges or patterns. In most cases, potentially bad data would need to be manually checked to confirm or reject the basis for the flag. Thus, a method is proposed to flag suspect data with adjustable scrutiny levels. Individual users may wish to apply higher or lower scrutiny based on their knowledge of the dataset or gained experience with previous data flagging (e.g. if most flagged data is determined to be valid data, the user may wish to lower the level of scrutiny, flagging only data further outside the expected range). Data quality check methods, developed based on empirical counts in BikePed Portal, are proposed to identify appropriate flags for repeated zero values, repeated non-zero values, and maximum/excessive count values. Finally, we developed recommended check thresholds along with an implementation approach and plan to incorporate additional data quality and control checks into BikePed Portal.


This is a final report, NITC-RR-1026, from the NITC program of TREC at Portland State University, and can be found online at:



Persistent Identifier