Palm distribution of road conditions and its application in incident risk analysis on main roads in Finland

This study was designed to apply the method proposed in two earlier studies conducted for the Ring road I in Helsinki to identify conditions when the risk of a traffic incident is elevated and develop the method further to be able to analyse the riskiness over whole country. We draw the following conclusions from this study: 1. The method developed earlier can be extended to all main roads without meeting computational difficulties when working with Mathematica. The key design principle was to build the Palm distributions per road segment and storing them in one file per road. 2. Because relatively few segments are measured continuously, we could not use actual traffic densities per hour. Instead, these were replaced by estimates using traffic variation factors. The results were throughout credible and the identified risky conditions were similar to the findings of the Ring road I study. 3. The analysis revealed road segments with very high risk levels. These top-20 lists should be studied by domain experts. Note that there may also be errors caused by the volume estimation method. 4. The overall road condition indicators "yellow" and "red" turned out to correspond roughly to risk levels 1.5 and 2.5, respectively. Our method could be applied to tune the use of the colour indicators.