Massive GNSS data for road safety analysis: Comparing crash models for several Canadian cities and data sources

Area of Science:

• Transportation engineering
• Traffic safety analysis
• Data science in insurance

Background:

• Mobile sensors offer valuable transportation data, but costs limited previous studies.
• Usage-based insurance (UBI) enables driver monitoring and premium adjustment using surrogate safety measures (SSMs).
• Contextual link-level SSMs (e.g., traffic volume, speed) can enhance UBI discount calibration.

Purpose of the Study:

• To quantify the relationships between contextual SSMs and crash frequency.
• To validate previous findings on SSMs and crashes using smartphone data.
• To compare these relationships across three Canadian cities using UBI data.

Main Methods:

• Extracted SSMs from UBI data were compared with historical crash data using Spearman's Rank Correlation Coefficient.
• Spatial Bayesian crash models were developed and validated.
• Analyses were conducted using data from Quebec City, Montreal, and Ottawa.

Main Results:

• Observed correlations between SSMs and crash frequency generally mirrored previous study directions (e.g., braking, congestion, speed variation positively associated; mean speed negatively associated).
• Correlation strength was slightly higher in the UBI data.
• Crash models estimated using UBI data demonstrated consistency across the three cities.

Conclusions:

• Contextual safety measures derived from UBI data are reliable and consistent across different urban environments.
• This study demonstrates the practical application of contextual safety measures to enhance UBI programs.
• Findings support the large-scale implementation of UBI for improved traffic safety and insurance calibration.