SPATIAL MODELLING OF ROAD ACCIDENT DATA

Maher suggested that spatial autocorrelation between the mean accident frequencies at neighbouring sites may account for the apparent 'migration' of road accidents from treated to untreated sites, as observed by Boyle and Wright. In this paper, the data used by Boyle and Wright, together with a new data set from the City of Birmingham, are analysed to measure the extent of spatial autocorrelation present. Theoretical spatial models which can be used to represent road accident data are examined and compared with these data sets. The results obtained from the analysis of the accident data indicate that spatial autocorrelation does exist between the numbers of accidents at neighbouring sites on a road network. There is quite a large variation in the size of this correlation, between both different years, and different road networks. Some of this variation may be due to sampling fluctuations. The existence of spatial autocorrelation in some data sets indicates a need to develop and fit models which allow for this correlation to be present. Of the models discussed, those based on flow are the most realistic. However the two stage stochastic may be more practical, as there is difficulty in obtaining flow data of high enough quality. For the covering abstract of the conference see IRRD 851540.

• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/isbn/019853650X
• Corporate Authors:

  CLARENDON PRESS

  WALTON STREET
  OXFORD,   United Kingdom  OX2 6DP
• Authors:
  • LOVEDAY, J
  • JARRETT, D
• Conference:
  • MATHEMATICS IN TRANSPORT PLANNING AND CONTROL (INSTITUTE OF MATHEMATICS & ITS APPLICATIONS CONFERENCE SERIES 38)
• Publication Date: 1992

Language

• English

Media Info

• Features: References;
• Pagination: p. 433-46

Subject/Index Terms

• TRT Terms: Conferences; Crash rates; High risk locations; Improvements; Mathematical models; Stochastic processes; Traffic engineering; Traffic flow
• ITRD Terms: 1663: Accident black spot; 1612: Accident rate; 8525: Conference; 9108: Improvement; 6473: Mathematical model; 6574: Stochastic process; 655: Traffic engineering; 671: Traffic flow
• Subject Areas: Data and Information Technology; Operations and Traffic Management;

Filing Info

• Accession Number: 00631269
• Record Type: Publication
• Source Agency: Transport Research Laboratory
• ISBN: 0-19-853650-X
• Files: ITRD
• Created Date: Jul 1 1993 12:00AM