統計理論應用於交通安全分析與績效改善評估方法

Apply Statistics Theory into Traffic Safety Analysis and Evaluation Method of Performance Improvement

管政夫(Cheng-Fu Kuan) ； 楊麗玲(Li-Ling Yang)

《中華技術學院學報》 35期 (2006/12) Pp. 231-243

https://doi.org/10.7095/JCIT.200612.0231

摘要

交通事故之發生極具機率性，其原因甚為複雜，無論人為因素造成抑或自然 環境使然，或者歸過於車輛機械、幾何設計、縱橫斷面、交通流量、車流組成……等都與事故發生之機會有關，交通管理者以及警勤執法者莫不希望事故減少至最低程度，交通安全研究者則欲從事故之現場或者以往資料探索、判讀、解析以提供見解協助促進交通安全。然交通安全之分析是一門專業，交通安全改善與否不能單以事件發生次數增減加以論斷。由於交通安全分析主要目的在於瞭解安全癥結所在以作為改善策略參考，並經一段時日後對改善績效加以評估，基此不同之交通安全分析目的應有不同方法。 評估交通安全狀況，表達方式大都以百萬車公里或登記車輛數、人口數、汽油消耗量及所發生事故次數加以評估，屬短期性交通研究之一種，吾人僅能瞭解狀況之表徵，其內含不得而知，若能應用統計理論做些檢定、估計、變異等之分析，則應有較具深度之見解，其他如各種形式之迴歸，微分方程都是交通分析有力之工具，而各種之交通分析首要條件就是資料完整性，一般而言研究者大都使用次級資料(Secondary Data)，使用前需加以檢定以瞭解資料機率分配性質，因事故為機率性之事件並非服從均勻分配(uniform distribution)或單調增加(monotonic-increasing)、單調減少(monotonic-decreasing)，關於此，統計理論卻也提供我們一些方法。

關鍵字

統計量 ； 一因子 ； 檢定 ； 臨界值

並列摘要

The occurrence of traffic accidents is full of probability. Meanwhile, the reasons are much complicated, which can be ascribed to artificial factors, or the natural environment, or vehicle mechanism, geometry designs, cross sections, transportation flow, the composition of vehicle flow, ……, etc. The traffic management authorities and the law-enforced parties frequently hope to see the accidents being reduced to the lowest degree. The traffic safety researchers then want to provide solutions to enhance safety by working on the scene of accidents, or investigating, studying and analyzing previous data. However, the analysis of the transportation safety belongs to a professional field. Whether the transportation safety is improved or not can not be simply judged by reviewing the occurrence number of accidents. The main purpose of analyzing traffic safety is to understand the crux problems of safety and to propose improvement strategies and then to evaluate the improvement performance after a while. As a result, there should be different methods for different analysis purposes of traffic safety.
We mostly adopt the occurrence number based on million-car kilometer, or vehicle registration number, population, consumption of gasoline as the evaluation basis, which is regarded as a short-term traffic research. However, we only can understand the token of situations but to reach the depth of occurrence. If we can apply statistics theories to do examinations, estimate, variation analysis …… etc., more profound solutions should be furnished. Besides these, various forms of regression, differential calculus of equation are deemed as powerful tools of traffic analysis. Doing this, the completeness of data is the predominant term of various traffic analyses. Generally speaking, traffic researchers mostly utilize Secondary Data. Before sourcing this method, examining data to realize the characteristic of data probability distribution is mandatory, for the accident occurrence is an event of probability and can not be classified as either uniform distribution, or monotonic increasing or monotonic decreasing. From these, we realize that Statistics theory indeed provide valuable methods.

並列關鍵字

Statistics ； One factor ； Test ； Critical Value