Private Sector Data for Performance Management
Final Report

July 2011

CHAPTER 3. PRIVATE SECTOR DATA MARKETPLACE REVIEW

A state-of-the-practice review was conducted to gather information on:

• Products and services offered by private sector data providers.
• Public sector agency uses of the private sector data products and services.

Private Sector Data Products and Services

To gather information, survey questions were prepared and supplied to participants prior to a scheduled telephone call. The private sector data providers that participated were:

• AirSage
• American Trucking Research Institute (ATRI)
• INRIX
• NAVTEQ
• TomTom
• TrafficCast

Table 2 contains a summary of the information relating to the provision of private sector historical data. This chapter provides a discussion of the items summarized in the table, as well as some additional detail on data parameters that relate primarily to real-time information.

Coverage

As identified in the notes following the table, with the exception of TrafficCast, all providers indicated a national coverage capability on main roadways, typically down to the major arterial street level. This corresponds to Functional Class (FC) 3 roadways in the TMC location referencing system.

Map Matching

All of the companies provide their data mapped to some system that allows for the geographic identification of the roadway segment to which it applies. With the exception of ATRI, which uses mileposts, all providers utilize TMC as a "minimum common denominator." INRIX, NAVTEQ, and TomTom also use proprietary mapping which allows data to be mapped to segments at a finer (smaller) resolution than TMC.

Data Available

Table 2 lists several types of data which are available across the different providers. Speeds (S) and travel times (TT) are the predominant data provided, available from all the vendors as historical data. All vendors also state the provision of some type of quality (Q) data, although the specific quality indicators vary by provider. All vendors except ATRI state the availability of Incident (I) data. NAVTEQ indicated the capability of traffic volume data for a portion of the network. Volume data comes from a variety of data sources including fixed-point sensors installed and maintained by public agencies, camera counts, and probe vehicles. It should be noted that data availability on a per-lane basis is still in its infancy as a provider offering, although a number of respondents indicated that this possibility was in research and development.

Table 2. Summary of Private Sector Data Products and Services

	Airsage	ATRI	INRIX	NAVTEQ	TomTom	TrafficCast
Data Available (a)	S, TT, I, Q	S, TT, Q	S, TT, I, Q	S, TT, I, Q, V (portion of network)	S, TT, I, Q	S, TT, I, Q
Services Available (b)	D, A, PM	D, A, PM	D, A	D, A	D, A, PM	D, PM
Data Source (c)(d)	Cell phone, 911, traffic counts	GPS on commercial truck-only fleets	State installed sensors, commercial fleets, consumer GPS	State installed sensors, commercial fleets, consumer GPS.	Consumer GPS, Fleet GPS	State installed sensors, commercial fleets, consumer GPS, Bluetooth systems.
Aggregation Levels for Historical Usage	None; as captured	1 mile, 1 minute	15 -60 minutes	15 minutes	1 hour	15 minutes
Accuracy Checks Performed	Visual camera count, Probe vehicles.	Anomaly checking done, routines not disclosed.	Independently verified in large-scale testing.	Data checks prior to map matching. Comprehensive drive testing.	Data checks prior to map matching.	Simple-adjacent points compared, some clients doing accuracy checks.
Documented Quality Levels	None provided. Stated they meet Section 511 requirements.	None-burden is on receiver of data.	Accuracy above 95% Availability above 99.9	None provided.	None provided. Stated they can meet Section 511 requirements.	None provided. Stated they can meet Section 511 requirements.
Pricing	Specific pricing information not provided.	Specific pricing information not provided. Not for profit.	Full use open licensing is $800 per mile per year plus $200 per mile one-time setup fee. 25% discount on other roads purchased in conjunction.	Specific pricing information not provided.	Specific pricing information not provided.	Specific pricing information not provided.

NOTES:

1. Data Available: "S"=Speed, "TT"=Travel Time, "I"=Incidents, "Q"=Quality, "V"=Volumes
2. Services Available: "D"=Discrete Data (individual data points), "A"=Aggregate Data", "PM"=Performance Measures
3. National Coverage: Not listed in table. All providers indicated national coverage, except TrafficCast which is currently in urban areas
4. Map Matching: Not listed in table. All providers except ATRI indicated a minimum use of TMC. ATRI uses mileposts. INRIX, NAVTEQ and TomTom also use proprietary segmentation more detailed than TMC.

Services Available

Table 2 identifies a service of "D" for discrete data, which means each individual (discrete) data point passing a segment within a timeframe. When purchasing discrete data, a consumer would get all of the individual speed or travel time points within a section, within a timeframe, whereas they would only get one value under the purchase of aggregate data.

Data Sources

Providers are using an expansive range of data sources including GPS data from fleet vehicles, commercial devices, cell phone applications, fixed sensors installed and maintained by other agencies, fixed sensors installed and maintained by the data provider, and cell phone location. Each provider that responded essentially had a unique (to some degree) set of data sources. While there was some overlap, no responding provider utilized exactly the same data model as another provider.

Aggregation Level

The predominant service offering is refined or aggregated data. Aggregate data are available from all of the responding providers on a historical basis. What is different across the providers is the level of aggregation. ATRI provides the smallest level of aggregation, at 1 mile or 1-minute. Other providers vary from 15 to 60 minutes. In part, the differences are due to the wide variety of data sources.

Data Quality and Accuracy

Most providers did not disclose specific quality evaluation results or quality assurance algorithms. With the exception of ATRI which stated that none are performed, TrafficCast did state that a part of their general methodology included simple adjacent point comparison routines but also stated that more sophisticated methods were employed. INRIX, in part due to the comparisons performed by the I-95 Corridor Coalition, stated that their data has been verified in large-scale client testing.

INRIX explicitly stated the capability of meeting an availability of more than 99.9 percent and an accuracy of greater than 95 percent. While it applies to the real-time data aspect, all providers were also asked about their awareness of, and capability to meet, the information delivery requirements in the FHWA's Section 1201 Final Rule. Providers were aware of the requirements and expressed no concern over meeting them. With the exception of ATRI, the data parameters for accuracy and availability would appear to exceed the FHWA rule-making requirements.

Pricing

In general, the availability of pricing information was minimal. Most providers appear to negotiate each purchase individually. Pricing is tied to the usage of the data. If data are used for a single application, one price point is employed. If data are used for multiple applications, other price points are utilized. Providers also make a distinction between uses, such as modeling or O-D studies, and derivative products, such as summaries distributed to external sources. While the various price points were not disclosed, all providers stated they exist. The INRIX pricing provided in Table 2 is the complete package pricing for real-time data with multiple-use ability of the historical data. Reduced requirements and uses would result in lower price points.

Public Agency Consumers of Private Sector Data

Table 3 summarizes the information for the public agency consumers of private sector data. The public agencies included in this summary are:

• Wisconsin Department of Transportation
• Houston-Galveston Area Council
• Michigan Department of Transportation
• Texas Department of Transportation
• Maricopa Association of Governments

Service Purchased

All respondents in Table 3 indicated the purchase of historical data versus a real-time data feed. Again, it should be noted that many providers do offer two types of historical data for purchase. The first allows for the purchase of each discrete or individually recorded data point, such as all vehicles and their speeds within a segment and time of day. The other, more commonly used method is the purchase of aggregate data which would provide one data point (such as a speed or travel time) for each section and time interval. Respondents uniformly indicated the purchase of the aggregate data type.

Data Purchase

Respondents were fairly evenly split on the particular attributes of the data purchase. Both travel times and speeds were identified as a purchase of interest, as well as performance measures. The finding of performance measures was somewhat interesting as during the provider survey, most providers indicated that they felt the marketplace would want to calculate their own measures from the purchased data.

Applications

Responses from data consumers indicated that the purchased data were used primarily for planning purposes, such as performance monitoring/congestion mitigation, origin and destination studies, or traffic modeling validation/calibration. Most agencies were purchasing a one-year data set although the Michigan DOT response indicated that it was for a time period of five years.

Validation

Only Michigan DOT and Maricopa Association of Government identified techniques in use to validate the purchased data and only Michigan DOT had specific criteria in place to measure those validation activities. In terms of availability, those criteria mirrored the FHWA requirements for real-time applications, while the accuracy requirements seemed significantly more lenient than what other studies (I-95 Corridor Coalition) have reported was possible.

Table 3. Summary of Historical Data Consumer Survey Results

	Wisconsin DOT	HGAC	Michigan DOT	Texas DOT(d)	Phoenix MPO (MAG)
Status?	RFI	Purchased	Purchased	Purchased	Purchased
Service Purchased (a)	H	H	H	H	H
Aggregation Level	Hourly day-of-week averages	15 min	5 min	Hourly day-of-week averages	Weekday
Data Purchased (b)	S/TT, PM	S/TT	S/TT	S/TT, PM	PM
Applications (c)	PM, TM	PM, TM, OD	PM	PM	PM
Coverage	All arterials	Houston region	MI Freeways	Statewide TMC network	Region
Timeframe	1-2 years	1 year	5 years	2009	1 year
Validation Criteria	Not yet established	Not yet established	Avail >99.5% Acc less than +/- 10mph	None	Not yet established
Validation techniques	N/A	N/A	Probe, fixed point , re-id	None	Probe, fixed point.
Pricing (in thousands)	$80,000 (Est.)	$77,000	$200,000 per year	$28,000	Negotiating
Licensing	Multiple Use	Multiple Use	Single Use	Single Use	Multiple Use
Multi-Agency	Yes				Yes

NOTES:

1. Service Purchased: "H"=Historical, "RT"=Real-time
2. Data Purchased: "S/TT"=Speed or Travel Time", "PM"=Performance Measures
3. Applications: "PM"-Performance or Congestion Monitoring, "TM"=Traffic Model Validation or Calibration, "OD"=Origin-Destination Studies
4. See http://apps.dot.state.tx.us/apps/rider56/list.htm for published study results.

Pricing

Not all agencies responded with pricing information. Understanding the pricing models in play are difficult as each provider has different pricing models based on the application, type of data, and coverage area. This was confirmed in the responses from the data providers as none except INRIX would detail their pricing structure. There is little capability to infer standard pricing information that would be applicable over a large (multi-state or national) region from the responses received.

While not detailed in Table 3, responding agencies indicated that the use of purchased data was generally considered for the reasons of:

• Cost-effectiveness
• Faster turnaround
• Accuracy
• Availability

Overall, the respondents' stated advantages to private sector data was:

• Consistency of the data collection approach.
• Comprehensive coverage.
• Frequency of updates.

Consumers generally expressed satisfaction with their purchases, but this assessment is tempered by the fact that few validation efforts have taken place.

Summary

Overall, this state-of-the-practice review indicates a growing market for public sector acquisition and usage of private sector data. Private sector data providers are aware that this market has potential, are interested in developing new product offerings, and are looking to expand their customer base. Public agencies see a number of advantages to private sector data, but are still working out issues such as validation of the data and matching data to a consistent location referencing system.