Classification Trees and Association Rules for Exploratory Analysis of Powered Two-Wheeler Crashes

摘要

Aim of the study was the exploratory analysis of powered two-wheelers crashes in Italy inorder to: (1) detect interdependence among crash patterns as well as dissimilarities amongpatterns, that cannot be deduced in straightforward way from any query to the data base, (2)find out non-trivial and unsuspected relations in the data, and (3) provide insights for thedevelopment safety improvement strategies focused on PTWs. At this aim, explorativeanalysis of the data relative to the 254,575 crashes involving PTWs occurred in Italy in theperiod 2006-2008 was carried out by data mining techniques.Classification trees analysis and association rules analysis were performed. Treebasedmethods are non-linear and non-parametric data mining tools for supervisedclassification and regression problems. They don't require a priori probabilistic knowledgeabout the phenomena under studying, and no assumptions are necessary. Moreover, trees arecomputational feasible and consider conditional interactions among input data. Associationdiscovery is the identification of sets of items (i.e., crash patterns) that occur together in agiven event (i.e., a crash in our study) more often than they would if they were independentof each other. Thus, the association rule method can detect interdependence among crashcharacteristics.Results of the two analysis methods were consistent each other. Simultaneously, thetwo techniques present different characteristics which make their joint use complementary.Classification tree analysis provides a pictorial representation of the data and theirrelationships and is easily understandable. Association rule analysis provides morequantitative information, identifies the significant dependencies among all the singleattributes of the data base, and allows the evaluation of the statistically power of the rules bythe lift value. Analysis results showed that PTW crashes are strongly sensitive to severalroad, environment, and drivers attributes. The results of the analysis was successful inproviding useful insight about the PTW crash patterns and their relationships. Basing on theseresults, engineering countermeasures and policy initiatives to reduce PTW injuries andfatalities were singled out. The use of classification trees and association rules must not,however, be seen as an attempt to supplant other techniques, but as a complementary methodwhich can be integrated into other safety analyses.