請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/16821
標題: | 三車道以上路段車禍事故之空間相依性分析 - 以台北市為例 Spatial Dependence Analysis of Accidents in Road Sections with three or more lanes – A Case Study of Taipei City |
作者: | Wei-Chien Hsu 徐維謙 |
指導教授: | 許添本(Tien-Pen Hsu) |
關鍵字: | 空間相依性分析,空間自相關,空間迴歸模型,空間追蹤資料, Spatial Dependence Analysis,Spatial Autocorrelation,Spatial Regression Model,Spatial Panel Data Analysis, |
出版年 : | 2020 |
學位: | 碩士 |
摘要: | 鑒於過去文獻可知,同一路廊之路段或路口之事故發生具有類似的區位與特性,存在相互影響關係,空間相依性的產生往往是因為鄰近的路段具有類似的事故影響因子,而路段事故因子之影響力會隨著距離遞減,因此建置事故分析模型時需要考量空間因素。 本研究目的在於將空間相依性納入事故分析模型進行探討,採用台北市41條近年來取消第三車道禁行機車之三車道以上路段進行實證。首先,使用Moran’s I值與LISA值二指標來檢測空間自相關情形,並以空間計量模型探討事故之影響因子,比較傳統迴歸、空間落遲模型與空間誤差模型橫斷面分析之預測能力,其次使用空間追蹤資料之空間落遲與空間誤差模型進行分析,期望獲得橫斷面模型無法取得之訊息,並比較橫斷面分析與追蹤資料分析可能存在之差異。 實證結果顯示,實證區域事故資料之Moran’s I值均顯著大於0,說明其有正向之空間相依性,考量空間相依性之空間迴歸模型解釋能力優於一般傳統回歸模型,在空間迴歸模型中,空間落遲模型之估計優於空間誤差模型,而空間追蹤資料分析結果亦顯示空間落遲模型為較適之估計模型,與橫斷面資料分析結果一致。 It is well-known from the literature that accidents in segments or interactions along the same corridor would have similar location and characteristics. Spatial dependence mainly derives from the similar accident impact factors between adjacent road sections. The influence of the accident factor in road segment will decrease as distance increases. Therefore, spatial factors need to be considered when constructing an accident analysis model. This paper aims to incorporate spatial dependence into the accident analysis model. We use a database of 41 road sections with three or more lanes in Taipei City which have been canceling the driving restriction of motorcycles on third lane in recent years. Firstly, two indices for measuring spatial autocorrelation are considered, including (i) Moran’s I Index and (ii) LISA Index. Secondly, we compared the explanation power of three cross-sectional analysis model including (i) ordinary least squares (OLS), (ii) spatial lag model (SLM), and (iii) spatial error model (SEM). Then, a spital panel data analysis with SLM and SEM is applied to obtain the unmeasured information in cross-sectional analysis. According to the empirical results, we find that the Moran’s I value of accidents is greater than 0, implying a positive spatial autocorrelation among segments. Besides, the spatial regression model performs better that the traditional OLS model. In the comparison of cross-sectional analysis models, the SLM is more appropriate than SEM. Additionally, in the spatial panel data analysis, results show that fixed-effect SLM is also the more appropriate model, which is consistent with the cross-sectional analysis. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/16821 |
DOI: | 10.6342/NTU202002758 |
全文授權: | 未授權 |
顯示於系所單位: | 土木工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
U0001-1008202003212300.pdf 目前未授權公開取用 | 7.94 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。