TRAFFIC ACCIDENTS IN URBAN AREAS. 3. FORECASTING MODEL FOR TRAFFIC ACCIDENTS AT SIGNAL-REGULATED FOUR-ARM INTERSECTIONS

The purpose of this study was to create a model for the forecasting of motor vehicle accidents at signal-regulated four-arm intersections. An attempt was made to quantify the different types of potential conflict and to find the relationship between these quantities and traffic accidents. The quantitative expression for a number of potential conflicts of a certain type is termed risk volume. The data for each type of conflict have been processed separately with regard to risk volumes and accidents taking place. The statistical processing has been carried out according to two different methods, linear and isotone regression. As a dependent variable the number of accidents has been used and as independent variables the flows of actual traffic streams have been used. The reliability of the models has been tested at 25 intersections in Gothenburg. The forecasting models gave 195 (linear) and 199 (isotone) accidents as opposed to the 179 that actually occurred. /TRRL/