A model for correcting road traffic and road risk data for local effects has been developed within a working group of the French Ministry of Transportation, including ONISR (the National Observatory for Road Safety), SETRA (the Technical Service for Roads Studies) and INRETS. Among all determinants, climatic conditions and the time of year have the major effect on daily data, and a method for monitoring the risk indicators by taking account of these local effects has been implemented on French data. The calendar effects appear around bank holidays on one hand, and at periods of major traffic movements related to the vacation periods on the other hand. About forty different types of days are concerned, which can be classified in three major classes of days. As for the climate effects, they are measured by the means of three variables (temperature, rain and frost), as a difference to an average daily effect, so that they can be neglected for prediction purposes. The author presents some results obtained for the 1985-1999 period, for each different indicator (the traffic volume - the number of vehicle-kms, the number of accidents and the number of victims, on main roads and on motorways), and analyses these indicators' evolution over the period. He shows how this method can be transferable to other fields, which are influenced by such local effects and in which daily data are obtainable.
Abstract