U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000
Third-party traffic data refers to traffic data collected by entities without following the TMG, AASHTO Guidelines for Traffic Data Programs (AASHTO 2009), and State highway agency established written procedures. Such data often rely on alternative technologies, data analytics, and modeling under the name of big data and big data analytics. When a Federal, State, or local transportation agency is planning to acquire such data for its transportation program and project development, the agency should take a comprehensive review on: (a) data needs within the agency's different offices, (b) data quality, (c) data ownership, and (d) data cost.
An agency's data needs are the drivers for collecting and acquiring traffic and travel data. Before considering acquiring any third-party data, it is critical to assess what data the agency currently possesses and understand how such data are collected, stored, and disseminated. No further efforts regarding third-party data should occur until such an inventory is done.
There are three scenarios that may be encountered during the inventory process. The first scenario is that the agency does not possess the needed data. The second case is that the agency does have the data but would like to determine whether the third-party data is a more effective route for acquiring such data. The third scenario is that the agency does have the data and is not planning to pursue any further with the third-party data.
When faced with scenarios one and two, the agency should coordinate with other agency units and offices (e.g., planning's modeling and forecasting office, NEPA and project development offices, design offices, asset management offices, freight program offices, operations, ITS, Federal-aid office, construction office for maintenance of traffic decisions) to refine the needs of the data regarding the exact nature of the data item. An
understanding of the needs for each unit, office, and/or program for a specified data item and how often this data item is needed will put the data collection and acquisition on a solid path forward.
Remember, data needs are the main driver for all other actions. The goal of this coordination step is to ensure that only the needed data are acquired. Once the data are acquired, they can and will meet expectations and will be fully utilized by all necessary offices and units, achieving the goals of acquiring the data properly once and using them multiple times by multiple users.
Given that the data being considered for acquisition is out of the agency's jurisdiction regarding how they are obtained, it is critical for the acquiring agency to understand and know the quality aspects of the data. Both data accuracy and biases are the two common parameters used to characterize data quality.
An agency representative, at a minimum, should ask the potential vendors about the following aspects regarding their data quality:
Additionally, more specific questions such as these listed below may also be inquired with the vendors:
In addition to inquiring about the data quality with the vendors, the agency should conduct its own data quality testing and review. Data quality testing and reviewing should rely on sound statistical methodologies, be systematic, and follow established traffic engineering methods wherever possible.
Lastly, on the data quality front, data to be acquired from a third-party vendor should be delivered in a format compatible with the agency's current software usages. The agency should avoid allowing the third-party data vendor's software as the only software capable of handling the data.
Acquiring third-party traffic data could be carried out with many different contractual specifications. Agencies should be fully versed with the implications of different contractual terms which may limit the agency's rights for the data. Specifically, the agency should pay attention to the items listed below.
Will the agency have the data in its own system, or will the data stay in the vendor's system? Are the data going to be transferred to the agency or the agency's designated recipient? How will the agency integrate the third-party traffic data with its current traffic database? These are very important decisions that need to be made. If the data are going to reside at the platform of the third-party data supplier, questions related to how the agency is going to access the data, how many people in the agency can access the data at the vendor's platform, and how many years such data will be residing on the vendor's platform should be specified in writing. In addition, when the contracting time expires, how the agency retains these data should also be specified, thereby preventing the agency from losing the data due to time.
If the agency is going to store the acquired data on its platform, make sure such data are acquired in a compatible data format. This is especially true if the data are being acquired for roadway link (segments of roadways) based data (e.g., AADT for Route 100 from Major Avenue to Capitol Boulevard). These links should be compatible with the agency's roadway geospatial network. In other words, link-based data must be readily conflated to the agency's highway network. Many agencies are finding proper conflation of datasets to be of significant concern due to possible dynamic assignment of traffic data onto incorrect segments in the host agency's network. Exercising caution and proper oversight of this element when acquiring data is most helpful to having a successful and useful product.
The ownership of the acquired third-party data should be clearly stated. With the specification of ownership, other terms need to be specified regarding the freedom of using such data for the agency's purposes such as publishing, sharing, and releasing the data.
The agency should establish the ownership rights for all data to be acquired and fully understand any unlimited rights to use, share, publish, and release such data without further consent from the third-party vendor or compensation to the third-party vendor in perpetuity. The agency may need to establish sharing limitations (i.e., share the data with other local agencies, but not with private parties). The interested agency should avoid the practice of renting or leasing data from the third-party supplier.
Version control refers to updating the same data produced earlier. Each dataset delivered should have a version specification so the agency can properly reference back to the supplier of the data in the future.
Cost is a significant factor in any agency's decision-making process regarding data acquisition. Understanding both the total cost of the vendor's product and any ongoing expenses is key to understanding the vendor's full cost. For example, will new software licenses be required? Will the data be acquired on an ad hoc (as needed) or a reoccurring basis (multiple times, multiple years)? What will be the cost of additional licenses if they are needed? The agency also needs to consider the costs from having agency staff or consultants assist in managing the data program.
Given the data acquired are typically well-defined, agencies are encouraged to use a firm fixed price approach in acquiring such data. Agencies should avoid negotiation for software packages used to handle such data as part of the data acquisition package. Data delivered by the third-party should avoid the situation where only the third-party data supplier software can handle the data.
When an agency is planning to acquire third-party traffic data, considerations to concerns and factors from four areas should be exercised. These areas are data needs, data quality, data ownership, and data cost. Begin by clarifying the data needs. The data needs are the drivers for collecting and acquiring traffic and travel data. Then, it is critical for the acquiring agency to understand and know the quality aspect of the data. Next, the agency needs to be clear on the data ownership. Under data ownership, the acquiring agency needs to be fully versed with the implications of different contractual terms which may limit distribution rights for the data. Lastly, but equally important, is the data cost. Understanding both the total cost of the vendor's product and any ongoing expenses are key to understanding the whole agency cost.
