Developing a spatio-temporal interactions model for car crashes using a novel data-driven AHP-TOPSIS

We have proposed a data-driven method for spatio-temporal analysis of car crashes based on the Multi Criteria Decision Making (MCDM) procedure in the Zanjan city, NW Iran which are recorded in the period 2019-2020. A combination of Analytic Hierarchy Process (AHP) and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is used in this paper to accurately identify the spatio-temporal interactions exist in car crashes. A data-driven AHP-TOPSIS procedure is arranged based on assigning proper weights to the time series related to the peak of accidents. On the other hand, for spatial analysis, the Kernel Density Estimation method was used to create the continuous-value maps of different traffic accidents and then classified using Natural Breaks Classifier. In fact, the proposed methodology can be used to identify car crash hotspots by considering spatio-temporal interactions as well as addressing exaggerated weightings arising from knowledge-driven modeling. By using the spatio-temporal interaction maps in which the location and time of crashes are considered, simultaneously, it is possible to provide a new scientific strategy for identifying car crash hotspots which can lead to better traffic management, improved allocation of resources, and enhanced prevention regulations.

Automated summary

This study proposes a data-driven method for spatio-temporal analysis of car crashes in Zanjan city, NW Iran. The combination of AHP and TOPSIS is used to identify the spatio-temporal interactions in car crashes, and a data-driven AHP-TOPSIS procedure is arranged based on assigning proper weights to the time series. The study also uses Kernel Density Estimation method for spatial analysis, providing a new scientific strategy for identifying car crash hotspots.