Developing speed-related safety performance indicators from floating car data

In the road traffic safety domain there is a need for using proactive (non-crash-based) indicators, known as safety performance indicators (SPIs). Traffic speed based on big data (floating car data [FCD]) could help develop network-wide SPIs, but related knowledge and experience are insufficient so far. The authors attempted to fill this gap by using nationwide Italian FCD to develop speed-related SPIs and validating their relationship to crashes to see their potential explanatory value. The authors calculated the coefficient of variance (CV), congestion index (CI), and the number of incidents as candidate SPIs. For validation, the authors used linear correlation, crash frequency model, and ranking consistency. Incidents turned out to be the best SPI, especially for motorways.