台灣都市土地價格空間聚集分析 －以1997~2006年為例

Urban Land Price Spatial Cluster Analysis in Taiwan－1997~2006 Case Study

鄭文馨(Wen-Hsin Cheng)

指導教授 ： 李泳龍

長榮大學/管理學院/土地管理與開發研究所/碩士(2007年)

摘要

傳統地價變遷模型係以線性與非線性迴歸模式分析，較忽略地理空間變動情形，無法反映地價空間分佈之區域變化情形。因此，本研究選取1997年至2006年台灣都市住宅區土地地價資料進行實證分析，運用空間統計方法，並配合地理資訊系統之展示功能，分別探討地價變動之空間相依性與空間異質性，並藉以解釋地價時空分佈，嘗試改善傳統靜態模型。由地價空間自我相關實證結果發現，台灣都市土地地價存在空間聚集之現象，顯示鄰近地區地價會影響當地地價，且區域地價分佈呈現南北兩極化之現象。此外，台灣東部與西部區域地價空間結構明顯不同；同時再由地價異質性實證結果顯示，影響地價之變數存在空間不穩定性，造成模式誤差項之空間自相關，進一步透過地理加權迴歸解決空間自相關問題，模式配適度檢定結果優於傳統最小平方法模型(OLS)，能符合都市土地價格實際空間變化。

關鍵字

土地價格 ； 空間統計 ； 空間自相關 ； 地理加權迴歸

並列摘要

The changing of land price of traditionally statistical methods was investigated by using linear or non-linear regression models. Owing to the geographically spatial variability was frequently ignored, it was unlikely to reflect spatial variability of the distribution of the regional land price. Thus, in this study, we take the Taiwan region as an example, by choosing the data of urban land price distribution to activate the empirical analysis during 1997-2006. This study is to explore spatial variability of the spatial autocorrelation and the spatial heterogeneity separately, by using methods spatial statistics combined the assisted with GIS analysis functions and by explaining spatio-temporal analysis of urban land price to try to improve traditional static model. The result of the empirical study of the spatial autocorrelation of urban land price shows that the urban land price in the Taiwan region has the tendency of spatial clustering. This demonstrates that land price of the neighborhood could influence local land price. Also, the distribution of the regional land price reveals huge gaps between the region of North and South of Taiwan. Besides, there is also an obvious discrepancy between the region of East and West of Taiwan in terms of the regional land price construction. In addition, the result of the spatial heterogeneity of the urban land price shows that variables influencing on urban land price do exist on Spatial Non-Stationarity, and this will lead to the problem of model residual of spatial autocorrelation. By applying spatial statistic of GWR, it not only solves this problem, also it reveals that the value of is more significant than this in OLS model. This outcome may help fine-tune to estimate land price which fitted in with changing of the real space.

並列關鍵字

Land Price ； Spatial Statistics ； Spatial Autocorrelation ； Geographical Weight Regression

參考文獻

36.賴碧瑩(2002)。經濟結構轉變後之地價變動分析，台灣土地研究，第五期，頁1∼頁22。

35.鄒克萬、黃書偉(2003)，台灣地區所得空間公平性變遷之研究，規劃學報，第三十期，頁47∼頁57。

14.胡海豐(2002)。住宅面積與環境寧適需求之取捨，建築學報，第三十九期，頁51∼頁62。

29.張齡方、古建廷、林俊男(2006)，以地理加權迴歸分析建立灌溉率與各影響因子之關係，農業工程學報，第五十二卷第二期，頁73∼頁82。

30.張齡方、古建廷、林俊男(2006)，以地理加權迴歸分析建立灌溉率與各因子之關係，農業工程學報，第五十二卷第二期，頁73∼頁82。