Identification of hotspots on roads using continual variance analysis

This paper presents a new statistical model for the identification of dangerous locations (subsections) on roads, also known as hotspots. The model is based on continual analysis of variance. The variance parameter has the potential for the synthesis of quantity and quality, especially regarding traffic accident frequencies and the consequences of traffic accidents within subsections and the significant comparison of the produced synthesis. The sensitivity of the suggested model can be adjusted with the level of disjunction and the length of subsections. A practical application of the new model is performed using a sample of 8442 traffic accidents, of which 6079 were Property Damage Only (PDO) accidents, 2041 resulted in injuries and 322 resulted in fatalities. The sample is for the period of 2001 to 2011 and is from an ‘I class’ two lane rural state road in the Serbia with total length of 284 kilometres. The results acquired using the continual analysis of variance were compared with previous results from four HotSpot Identification Methods (HSID) that are also based on the frequency of traffic accidents. First published online 12 April 201