Publikacje

Abstract:

This paper briefly presents our original approach to the problem of generating accurate and interpretable collections of linguistic, fuzzy rules from data using multi-objective genetic optimization method. The main goal of the paper, however, is the application of our approach to real-life road traffic collision data collected by the U.S. Department of Transportation to discover mechanisms and patterns governing the road accidents. 3591 records describing road traffic collisions in one of the biggest U.S. states, i.e., in California are considered. Each record is described by 25 input attributes grouped into 3 sets, namely, accident specific ones, road specific ones, and environment related ones. Our approach generated as many as 37 solutions to the considered problem; these solutions are characterized by different levels of compromise between the accuracy and interpretability of the obtained rule-based systems. The obtained results are the point of departure for further research including, in particular, the road safety data from Europe.

POWRÓT

Strona główna > Publikacje

Fuzzy modelling of road traffic collision patterns using multi-objective genetic optimization method