商業區人行系統行人流動特性研析與行為模式-以台北市信義商圈路口為例

Abstract

本研究選取台北市信義區商圈路口人行設施作為觀測地點，區分尖峰及離峰時段行人，參考引用車流相關理論及行人移動行為之相關文獻作為分析行人流動特性之依據。本研究於巨觀行人流動特性部分，分別探討行人流時段特性、流量比特性與QKV特性。另外，依觀察將行人行為細分為三類，超越前方行人、僅作橫向偏移與跟隨前方行人，考慮前後行人速度差異、跟隨間距、前方總瞬時密度、性別與性別異同等影響變因參數，構建判別函數與多項羅吉斯迴歸模式，用以解釋所觀察到之行為表現。結果發現，性別變數無論尖峰或離峰時段均不會影響行人行為，而性別異同變數僅在尖峰時段對於行人行為有顯著影響，且研究所構建之模式在判別分類以及行為預測上均可達85%以上之整體準確率。

Pedestrian flow is investigated in a intersection in Xin-Yi district of Taipei city. The samples are classified into two categories, rush hour and off-hour. The pedestrian flow characteristics of phase, flow ratio and QKV theory are discussed in the study. Besides, three types of pedestrian behaviors are analyzed: overtaking, lateral moving, and following. With the consideration of pedestrian speed difference, following distance, instant density, gender and difference of gender, the discriminant functions and multinomial logistic regression models are constructed. The result indicates that gender has no influence on the pedestrian behavior and the difference of gender is a significant variable only in the rush hour. The hit rates of classification and prediction are both higher than 85% in the study.