新竹市科學園區從業員工通勤旅次起迄分布及抽樣分析

Abstract

起迄資料的蒐集一直是交通規劃重要的一環，對於運輸規劃、管理、控制相當重要，傳統起迄資料收集包括車流量調查、家庭訪問等，相當耗費人力與物力，故國內外學者研究以蒐集最少資料，經由模式推估而得旅次起迄資料，但其推估的精確性則與模式及調查樣本抽取地點等關係影響甚大；本研究不同於傳統以路口抽樣調查蒐集起迄資料，首先本研究蒐集超過60%的新竹市科學園區從業員工住家區位及工作區位座標點對的資料來建立起迄表，並利用GIS將其居住區位的點座標分佈型態視覺化與道路及村里圖層套疊，提供決策者較為準確地了解新竹市科學園區員工居住於新竹市的分布狀態，能較為直覺地做為決策的依據。接著我們建立以最小化分區間旅次數變異數為目標式，考量分區形狀完整性等因素的起迄分區模式，利用階層分群演算法求解起迄分區模式，並建立新竹市科學園區從業員工通勤旅次起迄表，進行不完整資料差補，再放大至母體總數，最後透過K-S檢定，檢定不同抽樣百分比型態是否相同，以及多少抽樣百分比足以代表母體。

The Origin and Destination (O-D) table is one of the important elements for transportation planning, management and controlling. The core of this study is to establish Origin-Destination (O-D) table of Hsinchu Science Park (HSP) commuter trips, and to understand the distribution of HSP commuter trip O-D ends. Generally, the O-D table can be obtained by direct measurements, interviews, or traffic counts, followed by various models correspondingly to convert sample O-D table to the total O-D table. It requires tremendous budget and time, and dependents upon the accuracy of the input data like the sampling size of survey, locations of traffic counting points. Differ from usual, the study firstly collects the residence and workplace coordinates (HSP commuter trips O-D pairs) as a data set, which is converted from addresses of above 60% of Hsinchu Science Park (HSP) commuters who live in Hsinchu and work in HSP. Then the geocoded data is displayed with GIS, a two-dimensional visualization technique, to show the geographical distribution and provide intuitive sense of HSP commuter O-D distribution to decision makers. Second, this paper builds equal-trip-size oriented Origin and Destination zoning models respectively with zoning shape constraints. Then, this paper uses hierarchical aggregation procedure to solve the zoning models. Third, this paper establishes O-D table of the HSP commuter trips. Fourth, this paper uses improved D-S method to complete the missing value in O-D table. Then, this paper enlarges the sample proportionally to population size and completes the O-D table. Finally, this paper conducts the K-S test to determine if samples and population differ significantly, and to know how many percentages of samples can represent population.