EconPapers    
Economics at your fingertips  
 

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

Meysam Effati (), Jean-Claude Thill () and Shahin Shabani ()

Journal of Geographical Systems, 2015, vol. 17, issue 2, 107-135

Abstract: The contention of this paper is that many social science research problems are too “wicked” to be suitably studied using conventional statistical and regression-based methods of data analysis. This paper argues that an integrated geospatial approach based on methods of machine learning is well suited to this purpose. Recognizing the intrinsic wickedness of traffic safety issues, such approach is used to unravel the complexity of traffic crash severity on highway corridors as an example of such problems. The support vector machine (SVM) and coactive neuro-fuzzy inference system (CANFIS) algorithms are tested as inferential engines to predict crash severity and uncover spatial and non-spatial factors that systematically relate to crash severity, while a sensitivity analysis is conducted to determine the relative influence of crash severity factors. Different specifications of the two methods are implemented, trained, and evaluated against crash events recorded over a 4-year period on a regional highway corridor in Northern Iran. Overall, the SVM model outperforms CANFIS by a notable margin. The combined use of spatial analysis and artificial intelligence is effective at identifying leading factors of crash severity, while explicitly accounting for spatial dependence and spatial heterogeneity effects. Thanks to the demonstrated effectiveness of a sensitivity analysis, this approach produces comprehensive results that are consistent with existing traffic safety theories and supports the prioritization of effective safety measures that are geographically targeted and behaviorally sound on regional highway corridors. Copyright Springer-Verlag Berlin Heidelberg 2015

Keywords: Spatial analysis; Machine learning; Road safety; Crash severity; Spatial dependence; Spatial heterogeneity; Wicked problems; C14; C45 (search for similar items in EconPapers)
Date: 2015
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (1)

Downloads: (external link)
http://hdl.handle.net/10.1007/s10109-015-0210-x (text/html)
Access to full text is restricted to subscribers.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:kap:jgeosy:v:17:y:2015:i:2:p:107-135

Ordering information: This journal article can be ordered from
http://www.springer. ... ce/journal/10109/PS2

DOI: 10.1007/s10109-015-0210-x

Access Statistics for this article

Journal of Geographical Systems is currently edited by Manfred M. Fischer and Antonio Páez

More articles in Journal of Geographical Systems from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().

 
Page updated 2025-03-19
Handle: RePEc:kap:jgeosy:v:17:y:2015:i:2:p:107-135