Using data mining in trip distribution modelling

Abstract

This scientific project, which will be realized in a period of 12 months, describes the main objectives and methodological procedures to be performed. The key objective is to investigate, through data mining technique (Decision Tree) and disaggregated data, the main variables that influence the trip distribution modelling. Furthermore, it is expected to compare the variables identified by the Decision Tree model and variables commonly used in traditional trip distribution models. The database to be used corresponds to Origin / Destination Survey, performed in 2007 in São Paulo Metropolitan Area. Aggregated and disaggregated variables will be used to investigate the choice of destinations (displacement patterns). The method to be developed includes the steps of ( 1 ) literature review , (2 ) data treatment, (3 ) grouping of traffic zones , (4) application of data mining technique , ( 5 ) Comparison of variables obtained with the usual trip distribution models and data mining analysis.