Author(s)

LEON PROCHOWSKI, HANNA KOCHANEK, MIROSŁAW GIDLEWSKI, TOMASZ PUSTY

Abstract

Numerical data about road accidents, collected in monthly steps, were used to build a time series. The series, covering the years 2010–2016 and 16 regions, was analysed to identify some time-dependent and regional circumstances that cause high variability in the number of accidents in annual cycles. In consideration of the nature of these data, they have been described with taking into account the trends as well as random and cyclic variations related to the annually repeating changes. In the process of clearing the time series of the trend component, a multiplicative model of seasonal variations was used. As a result, a time series was obtained, which exclusively consisted of seasonal and random variations. The time series became a basis for the carrying out of medium-term and short-term analyses, which covered seasonal fluctuations in the number of accidents. As an objective, the regional factors having an impact on the course of individual phases of these variations were sought. The like phases of the cycles were identified and then their numerical characteristics were compared with each other and referred to specific regions of Poland. The research results provide grounds for indicating the factors that generate a relation between the cycles and the territorial distribution of accidents. The knowledge of the factors that are important for the occurrence of road accidents enables the forecasting where and when seasonal actions should be taken in the field of e.g. road traffic organization, work of traffic control and road rescue services, etc. Thanks to the data analysis method adopted, the impact of seasonal variations on the course of changes in the regional accident hazard could be isolated.

Keywords

road accidents, seasonality index, annual cycles, territorial distribution of accidents