Behavioural factors in accidents at road junctions: the use of a genetic algorithm to extract descriptive rules from...

In-depth studies of behavioural factors in road accidents using conventional methods are often inconclusive and costly. To explore an alternative approach, 200 cross-flow junction road accidents were sampled from the files of Nottinghamshire Constabulary, England, coded for computer analysis using a specially devised 'Traffic Related Action Analysis Language', and then examined using a genetic algorithm (or 'rule-finder' program). For comparison, the same analyses were carried out on 100 descriptions of safe turns, and 100 of hypothetical accident descriptions provided by experienced drivers. A number of findings emerged, distinguishing between accidents turning onto and off major roads; those of younger and older drivers; those that resulted in injury or in damage only; and so on. The study suggests that multiple case-studies based on police case files could be very promising, provided homogeneous samples of accidents are used. Genetic algorithms could play a useful role in preliminary assessment, but an 'information bottleneck' remains to be overcome in capturing enough detail from each case for analysis.