基隆市交通肇事分析及安全改善之研究

Abstract

本研究以基隆市轄區民國89年至91年間所發生之肇事案件為分析對象，將肇事地點區分為路口與路段二部份，先以因子分析中的主成分分析法整併相關肇事因素，再以一般化線性模式中的卜瓦松迴歸模式及負二項迴歸模式分別構建基隆市路口及路段之肇事分析模式，結果發現基隆市路口及路段之肇事模式皆以負二項模式較為適合。由肇事因果分析關係發現，在路口部分，以能見度因子、交通量因子、路面狀況因子、道路型態因子、視線因子、號誌因子等六個變數與肇事次數呈現顯著相關；而在路段部分，則以路面狀況因子、視線因子、道路限制因子、道路型態因子、分向設施因子等五個變數與肇事次數呈顯著相關，惟能見度因子與肇事次數之關係較不顯著。由基隆市前十大高肇事路口及路段之肇事型態與肇事模式變數進行比較，發現各路口及路段肇事型態與本研究模式中各因子之正相關變數大致相符。 本研究另亦針對各高肇事路口及路段之肇事型態特性分別進行成因分析並研擬安全改善對策。在路口方面係以夜間及假日高肇事特性之路口為對象，在路段方面則以直路、彎道及夜間高肇事特性之路段為對象，以期提供有關單位改善交通安全及後續研究之參考。

This study is based on the injury and fatal accident data for the period 2000-2002, to analyze the characteristics of accidents for Keeling City.The principal components analysis of factor analysis is used to integrate the possible influenced factors of accident separately for intersections and road sections. Poisson regression and Negative binomal regression methods of generalized linear models are then used to construct the accident analysis model for intersections and road sections. The results show that Negative binomal regression model can well describe the intersections as well as the road sections. Further, the results show that intersection-related accident factors are visibility factor, traffic flow factor, road condition factor, road type factor, sight distance factor and signal factor. On other hand, the road section-related accident factors are road condition factor, sight distance factor, road restriction factor, road type factor and median treatment factor. Comparing with the accident types of the top 10 highest accident intersections and road sections, the positive related factors of accident models are very close to the real data. In this study, possible causes of accidents and correction measures to improve safety are also identified and proposed for selected high accident locations. For intersection-related accidents, it is focused on the night and holiday accident; for road section-related accidents, it is focused on the straight road, curve and night accident.