Exploration and evaluation of crowdsourced probe-based Waze traffic speed

ABSTRACT

Real-time traffic information such as traffic speed is essential to a variety of practical applications. Because of the high installation and maintenance cost of convention data collecting methods, transportation engineers have sought alternative data sources to monitor traffic. Probe-based traffic data, such as Waze traffic data, could serve as alternative data sources in traffic management, but this source has not been thoroughly explored nor evaluated. Using the 10.8 mile stretch of I-40 in Knoxville, Tennessee, we compared the speed measurements from both Waze and Remote Traffic Microwave Sensors (RTMS) over two months and explored the characteristics of Waze traffic speed data. These are the main findings: 1) These two datasets showed a similar pattern with slight differences. Waze speeds tend to be higher than RTMS speeds for high speed, while Waze speeds are more likely to be similar or even lower than RTMS speeds for low speed; 2) several factors affecting the speed differences between RTMS speeds and Waze speeds were identified, such as Waze speed value, time of day (peak hour vs. non-peak hour), AADT (Annual Average Daily Traffic), and segment length; and 3) Waze reported the same speed for several successive reporting periods if the real-time speed is not available. The effective Waze sampling period is about two to four minutes. Waze speeds had more real-time speed observations during congested times, indicating that Waze speeds are more reliable for congested scenarios. The findings may lead to a better understanding of this source of data and any resulting analysis.

KEYWORDS:

• Traffic speed
• radar sensor
• Waze
• crowdsourced
• probe-based
• probe vehicle

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Tennessee Department of Transportation [RES2018-01]; Tennessee Department of Transportation [RES2019-11].