Geospatial and machine learning techniques for wicked social science problems: analysis of crash severity on a regional highway corridor

A semantic‐based classification and regression tree approach for modelling complex spatial rules in motor vehicle crashes domain

“…Therefore, they could barely provide a sharable and reusable capability for motor vehicle crash prediction process. Moreover, a data mining algorithm alone is incapable of establishing correspondences and interrelations among different domains of spatial entities and complex relations of crash contributing factors . Integrated use of ontologies and data mining in a geospatial domain could significantly improve modeling of complex spatial relations for motor vehicles crash severity prediction in highway corridors.…”

“…11,[14][15][16][17][18][19][20][21][22][23] Moreover, due to the importance of knowledge acquisition from large data set of crash records, data mining methods have become an essential component in many road safety studies. Numerous data mining 12,13,[24][25][26][27][28][29][30][31] related studies have been undertaken to analyze crash records locally and globally with frequently varying results depending on the study area and the methods used.…”

“…Moreover, a data mining algorithm alone is incapable of establishing correspondences and interrelations among different domains of spatial entities and complex relations of crash contributing factors. 13 Integrated use of ontologies and data mining in a geospatial domain could significantly improve modeling of complex spatial relations for motor vehicles crash severity prediction in highway corridors. Formal ontology has been used in the application of information systems 32 and geospatial information domain.…”

“…in road vicinities . Each technique is developed for a specific intercity road and needs to be redeveloped for other corridors of interest . In addition, they could not put into use the complex relations among crashes factors and crash severity for the proper prediction of accident‐prone vehicles …”

“…[8][9][10] Each technique is developed for a specific intercity road and needs to be redeveloped for other corridors of interest. [10][11][12][13] In addition, they could not put into use the complex relations among crashes factors and crash severity for the proper prediction of accident-prone vehicles. 8,10 The size of spatial databases and the complexity involved in dealing with spatial attributes of crashes require the application of specialized data mining and geospatial techniques.…”