Mining Characteristics of Speeding and Red Light Running Violations Using Association Rules

Abstract

Speeding and running red light always belong to the major traffic enforcement target for preventing accidents caused by violation. In the northern metropolitan area in Taiwan, speeding and running red light accounted for 12.56%and 12.97% of total violation citations from 2011 to 2015. This paper aims at analyzing features of such violation including violation area, day of week, time of day, gender, age, regulation, and mode of each violation record by using Apriori algorithm of association rules. Total 40 rules with lifts higher than onewere discovered for speeding and red light running drivers.It was found that drivers with violation of speeding and red light running have the odds ratios higher than one. That means they are easier being involved into accidents. The results show the characteristics of speeding and red light running drivers and give evidence of the risk of these violations.