Road traffic accident analyze aggregation in spatial and time Pavel Hrubeš, Ing. Ph.D. 23. 9. 2009
Accident reports and analyzes purposes Government, legal policy makers Road maintainers, traffic engineers Integrated rescue systems (police, fire dept., EMS) Space and time resolution Country/Union District/Region Municipality/ Local zone Years/decades Country target plans Country comparison District target plans Budget allocation District comparison Black spot identification/ Local measure Months and Weeks Policy makers, measure Season trends Preventive actions Days, Hours and Minutes Operation plans Czech Technical University in Prague - Faculty of Transportation Sciences Department of Control and Telematics
Space of analyzes - accident description Year, Quarter, Season, Month, Week, Weekday Day, etc. Easter, Holiday, Christmas GIS related information's Weather measurement (wind, rain, fog etc.) Special occasions (football matches, concerts ), Socio-demographics data (population, count of drivers, male, female, age of drivers etc. Road system data (statistical length of roads, curvature) Law and driving rules changes Car systems data (CAN) Type and Cause of Accident, Count of Deaths, seriously or slightly Injuries Car Type and make of car Alcohol, Seat belts Weather condition Sight situation Descriptive information Time Accident Coordinates, Country, State, Region, City, Village, etc. Road, Crossroad or Street Number Type of Road or Crossroad Direction of driving Location Czech Technical University in Prague - Faculty of Transportation Sciences Department of Control and Telematics
Methods of analyze and issues Aggregation Time and spatial inf. Density estimation, ranking Cluster analyze Spatial and descriptive inf. Estimation of accident frequency Time (spatial selection) Small area estimation problem/disease mapping problem Many attributes, no known relevance and significance to analyze How to compare, sort, present, etc.
Accident severity index ASIn Accident Severity Index of n-th accident K number of persons killed SI number of seriously injured SL number of slightly injured M material damage estimation W.. weight of component C scale constant … other components Traditional cost function Accident Cost= K * Kcost + SI * SIcost + SL * SLcost + …
Aggregation – usual functions Sum of ASIn Sum of several ASI is almost equal to accident count. Average of ASIn Average of ASI aggregation shows ratio of accident’s severity to the count of accidents. Multiplication of ASIn Multiplication of several ASI for light accidents is also close to the one and grows very slowly opposite to the multiplication of ASI for serious accidents which grows faster. Color scale of aggregated ASI good/better average/static poor/worse
Aggregation of ASI in Time
Aggregation of ASI in Time
Aggregation of ASI in Time
Aggregation of ASI in Time
Aggregation of ASI in Time
Aggregation of ASI in Time
Aggregation of ASI in Time Month SUM, AVG and Multiplication ASI 2008 Day in the Month
Aggregation of ASI in space
ASI computation and aggregation Conclusion ASI computation and aggregation Simply control of analyze by adding components and modification of its weights Very simply implementation in relational databases and GIS Summary of ASI enable simple selection, sorting and presentation ASI could be useful by other methods of accident analyze as the input. (weighted density estimation, etc.)
Thank you for attention. Acknowledgment I gladly thank to Col. Josef Tesařík and Petr Sobotka from Directorate of Traffic Police, Branch of the Police Presidium of the Czech Republic which kindly provide me the official road accident data for years 2007 and 2008. Ing. Pavel Hrubeš, Ph.D. Czech Technical University in Prague Faculty of Transportation Sciences Department of Control and Telematics Konviktska 20, Prague 1, 110 00 Czech Republic Web: http://www.lss.fd.cvut.cz/ Email: hrubes@lss.fd.cvut.cz Tel.: (+420) 224 359 629 Czech Technical University in Prague - Faculty of Transportation Sciences Department of Control and Telematics