Evaluating the Reliability, Coverage, and Added Value of Crowdsourced Traffic Incident Reports from Waze

Amin-Naseri, Mostafa; Chakraborty, Pranamesh; Sharma, Anuj; Gilbert, Stephen B.; Hong, Mingyi · 2018 · OpenAlex-citations

DOI: 10.1177/0361198118790619

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study evaluates the reliability, coverage, and added value of crowdsourced traffic incident reports from the smartphone application Waze, addressing the need for traffic managers to understand the validity of this emerging data source. While traditional Advanced Traffic Management Systems (ATMS) rely on sensors, cameras, and law enforcement, these methods are costly and often insufficient for full road monitoring, particularly in rural areas or during low-traffic hours. The research aims to quantitatively compare Waze data against existing ATMS records and third-party probe data to determine its potential as a complementary monitoring tool. The researchers analyzed one year of Waze data (2016) from the Iowa Department of Transportation, comparing it against validated incidents in Iowa’s ATMS and incident detections derived from INRIX probe data. The evaluation employed a four-step procedure: matching Waze reports with ATMS records, matching Waze with INRIX data, estimating false alarms using manually labeled traffic camera images from the Des Moines metropolitan area, and calculating the potential additional coverage Waze provides. A hybrid matching function utilizing temporal, geographic, and semantic criteria (such as road name, direction, and incident type) was used to reconcile data across sources. The findings indicate that Waze is a reliable and timely source of traffic information. Waze reports covered 43.2% of ATMS-recorded crashes and congestion incidents, ranking as the fourth most significant detection source in Iowa’s ATMS. Waze detected incidents on average 9.8 minutes earlier than INRIX probe-based detections. The false alarm rate was extremely low, at 0.3%, based on camera verification. However, Waze’s reliability varied by time; reports were significantly less frequent and reliable between midnight and 6 a.m. due to lower user activity. Logistic regression revealed that time of day and road type significantly influenced the likelihood of an incident being reported on Waze. Crucially, the study estimated that 34.1% of Waze’s crash and congestion reports represented potential incidents not captured by existing ATMS sources, highlighting substantial added coverage, particularly for congestion in non-metropolitan areas. The significance of this work lies in demonstrating that crowdsourced data can effectively complement traditional traffic monitoring systems. The study provides a flexible framework for evaluating social sensor data and suggests that Waze offers broad geographic coverage and timely detection that existing sensors miss. These findings support the integration of crowdsourced data into traffic management strategies, potentially reducing reliance on expensive infrastructure while improving incident detection coverage, especially in areas lacking dense sensor networks.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• naturalistic crash near crash