INRIX Volume Profiles (Trial)

INRIX Volume Profile is a high-quality dataset with nationwide coverage across 2.6 million miles of road in the USA, offering the most comprehensive data that includes vehicle volume by street direction, time of day and day of week.    

• Dayparted - INRIX Volume Profile is the only vehicle count database that provides the number of typical vehicles by time of day and day of week in 15-minute bins by road segment on over 2.65 million miles of US roads.  
• Directional - Right hand turns are easier than left hand turns for drivers. Thus, retailers prefer to be on the right side and understanding the vehicle count by direction is important to ensure the highest ROI of a location.  
• Current - The Vehicle count database is from 2019, unlike government sources which can be several years out of date.   NOTE: This free trial dataset contains Volume Profiles for a section of the Seattle Metro area.

The data is provided in .csv format, with the following columns:

• OSM segment ID (see OSM reference document included with the dataset for details)
• Time bin (96 15 minute bins per day -- 0 is midnight and 95 is 11:45 PM)
• Day (day of week)
• AADT (sum of the 96 bin volumes for the segment and day)
• Bin volume (vehicle count for the 15 minute time bin)

OSM Segment ID|Time Bin|Day|AADT|Bin Volume ----|----- 100951670_0|41|SUN|531|10 100951670_0|42|SUN|531|10 100951670_0|43|SUN|531|10 100951670_0|44|SUN|531|10 100951670_0|45|SUN|531|11 100951670_0|46|SUN|531|11 100951670_0|47|SUN|531|11

One of the primary advantages of INRIX Volume profiles is accurate and current day parting and directionality. These are represented internally as a “volume shape”, one that can be scaled based on the known statistical relationship between the number of vehicles crossing a road segment and the number of mobile GPS devices observed to cross the same segment. Developing these “volume shapes” is the first step in creating INRIX Volume Profiles. To devise volume shapes from GPS data, trip-like sequences of GPS points are identified such that each sequence reliably represents a trip (or a section of a trip) taken by one or more persons in a single motor vehicle. A series of methods are used to estimate the most likely path taken by each traveler along their trip, to identify the road sections that were crossed and the associated time of crossing. Thus, for each trip, a list of road segments and associated crossing times are obtained, which can be aggregated to estimate the mean crossing count by road segment, travel direction, day of the week, and time of day. After processing to remove errors and anomalies as well as some regional calibration, these “crossing count profiles” essentially represent a scaled-down version of a volume profile, where the scale is approximately equal to the penetration rate of the contributing mobile devices.  

Knowing that the GPS devices used to develop the volume shapes represent a fraction of all vehicles on the road, it is necessary to scale these shapes to represent the true vehicle volume for any given road. INRIX does this by first gathering vehicle population and / or volume from a variety of public and private sources. By comparing the observed (i.e. sampled) vehicle population with the true vehicle count at a set of locations, unique scaling factors can be developed for different spatial regions and road types. The mobile location data that form the basis for INRIX Volume Profiles comes from a wide variety of device types and software applications, each of which represents some small (and possibly not representative) sample of the overall population. The combination of all of these sources is expected to be much more representative of the population than any one in isolation. That said, the penetration rate of different combinations of data providers can change by region and even within different parts of a city. Thus, Volume Profile scaling factors are developed based on measured penetration rates on a sufficiently small spatial scale to ensure that local trends are represented, and that regional differences in sampling rate are effectively dealt with.