張國楨Chang, Kao-Chen曾露儀2019-08-292014-8-212019-08-292013http://etds.lib.ntnu.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=id=%22GN0699230317%22.&%22.id.&http://rportal.lib.ntnu.edu.tw:80/handle/20.500.12235/94397不同以往都市發展相關研究採用「土地利用」、「人口數/人口密度」作為都市發展程度進程的量測方式，本研究改以「第三級產業」的規模作為一可量測都市發展進程的指標，藉由工商普查資料中第三級產業場所單位面的「場所單位」、「全年薪資支出」、「就業員工人數」、「樓地板面積」、「全年生產總額」等五個變項，以主成分分析方法濃縮，並命名為「都市發展程度指標」。 　　本研究以網格100公尺為空間單元，將1991年、1996年、2001年、2006年臺北盆地淡水河系右岸各村里的都市發展程度指標網格化，進而輔以空間統計方法Getis-Ord Gi*觀察臺北盆地淡水河系右岸的都市發展空間變化，歸納四個年度的都市發展熱點空間，發現臺北盆地淡水河系右岸的都市發展是由西邊逐漸減弱而東邊日益增強，印證臺北市區自1999年起至2006年，都市發展進程主要為由西向東擴展。 　　接著本研究針對2001年的都市發展進程指標，以可及性要素（捷運站出口距離、主要道路距離、道路面積密度、公車班次）及生活機能要素（宅內人口數）作為因子，藉由全域線性迴歸及地理加權迴歸方法進行因素分析。最後比較兩種迴歸方法的結果，發現相較於使用全域線性迴歸方法，使用地理加權迴歸能使模式的解釋力R2由0.51提升至0.73，提高將近22%個解釋力。兩種模式的準確度也可藉由AICc指標來檢驗，地理加權迴歸AICc指數為27846.20，比多元線性迴歸AICc指數34547.81來得低，代表地理加權迴歸模式相較於多元線性迴歸可以得到較為準確的推估結果，且也能夠具體展現模式中，自變項因子係數解釋力的空間差異性，進而提供更多訊息以了解都市發展進程空間展演的可能原因。 　　最後，本研究藉由地理加權迴歸係數解釋力的空間差異性及殘差空間分布圖，發現臺北盆地的都市發展進程，有相當程度是受到都市土地使用分區及都市歷史發展脈絡的影響，例如商業區及住商混合區的都市發展程度相對來講較高；而臺北車站及忠孝東路一帶呈現高度發展原因，深受當初設立定位影響。In the past, we used the data of land use and population as city definition of city development. Now, by adopting a new approach, we have found a total different way to view and define city development. In this paper, we use the PCA statistic way to draw up the index, and the data of commerce and service to present city development as well. By using the grid device of commerce and service data of years of 1991、1996、2001、2006 at the basin of Taipei, right beside the Tamsui river. Using the spatial statistic way of Getis-Ord Gi*, we can see the four-years index of city development hotspots in Taipei basin, and we have found that the magnitude of this index had been weakened at the west side and had been strengthened at the east side of the Taipei city, which gives us a strong and solid evidence to confirm the truth that Taipei city had developed from west to east. Then, we use the distance of MRT exits, the distance of main roads, the density of road areas, and the number of runs of scheduled buses as the factors of OLS and GWR. We have found that using GWR could make the explanation of model’s adjust R2 raise from 0.51 to 0.73, which is nearly 22% promotion. Moreover, we can use the index of AICc to show that GWR have more accurate model outcome. The index of AICc of GWR is 27846.2, and the index of AICc of OLS is 34547.81, in which we can see that the AICc index of OLS is much larger than that of GWR. GWR can also provide the model’s factors explanations in spatial way, offering more messages to know the city development better. Finally, in this paper, we uses the explanation and residual error of GWR model to find that Taipei city development had been affected by the strategy of zoning and the history of city development. For example, the degree of city development of commerce district and residential and commercial mixed-used district is much higher. Also, and the orientation is the main reason why the degree of city development of train station and Zhongxiao road are much higher臺北都市發展地理加權迴歸主成分分析TaipeiCity DevelopmentGeographically Weighted RegressionPCA