Analysis of Types of Crashes at Signalized Intersections by Using Complete Crash Data and Tree-Based Regression

摘要: Many studies have shown that intersections are among the most dangerous locations of a roadway network. Therefore, there is need to understand factors contribute traffic crashes at such locations. One approach classify and quantify effects configuration, geometric characteristics, volume on number signalized intersections. This paper addresses different affect crashes, by type collision, It also looks into quality completeness crash data effect incomplete final results. Data from multiple sources were cross-checked ensure all including minor usually unreported or not coded databases. The tree-based regression methodology was adopted in this study cope with multicollinearity between variables, missing observations, fact true model form unknown. results showed significant discrepancy found collision types their influence each model. two differences comparison total as base case be models head-on left-turn crashes. important relatively consistent for rear-end, right-turn, sideswipe when considered. However, angle changes structure added set because these less stable. Finally, characteristics correlated