都市公園的寧適效益可資本化入周圍不動產的價格上，並在空間面向呈現其正的鄰里效應與地方外部性，然而此類寧適效益抽象而非具體，不易進行衡量，遺漏了將造成環境改善整體效益的偏誤。本研究目的運用地理加權迴歸模型空間迴歸分析（空間特徵模型Spatial Hedonic Models 之一）研究方法，在空間自相關與空間異質性之診斷及空間迴歸分析改進傳統特徵價格法，以台北市大安區都市公園系統為實證範圍，探討住宅房價影響各因素邊際作用大小及其空間分佈變化狀況，為都市公園在城市鄰里居住環境之空間面向所形成的寧適效益研究提供幫助。研究結果發現在樣本點300公尺半徑內隨大小層級組合不同之都市公園面積，公園面積越大，房地產價格增加率越大，都市公園面積與鄰里效應成正比關係。都市公園距離與鄰里效應成反比關係，不同層級都市公園各在不同距離範圍內，離公園越遠，房地產價格遞減。地理加權迴歸模型可以改進傳統空間迴歸方法，可對城市地價影響因素邊際價格作用空間變化性進行良好的估計，可提高迴歸之解釋能力，以更具體和直觀的特性來探討都市公園之鄰里效應所帶來的影響與效益。同時因為地理加權迴歸模型中融入了地理座標，可在不同樣本點的各種係數估計間進行對比和檢查，易於發現問題，並有利輸出結合於GIS的空間分析，以更具體和直觀的特性來探討都市公園之鄰里效應所帶來的影響與效益。
It is difficult to measure the amenity benefits of urban park. However, its amenity benefit will capitalize into the prices of the surrounding real properties and present the spatial dimensions of positive neighborhood effects, localized externalities and localized amenities benefits. The purpose is to explore the factors and their marginal effects that impact on the residential housing prices based on Geographically Weighted Regression (GWR) Model(one of Spatial Hedonic Models),which can be used as references for scientifically governing the spatial patterns of the amenity benefits of urban green neighborhood living environments. This study chooses the urban park system of Da-An district in Taipei city as the empirical area. The major findings are as follows: There is a positive correlation between urban park area and neighborhood effects. The combination of deferent level urban park area within a 300 meter radius around a property revealed that the size combination of deferent level urban park area has a positive amenity impact, the more area of urban park, the greater the change in housing prices. There is a negative correlation between the distance to urban park and neighborhood effects, the distance to urban park decreases housing prices far from urban park. The more distance to urban park, the greater the change in housing prices. The paper is concluded with that GWR- based model can improve the traditional space regression methods and can be used to effectively estimate the spatial trends of marginal effects of the factors that impact on urban residential land price. The explanation power of GWR model is higher than traditional hedonic price method, each coefficient estimates value appeared to estimate, propose analyzes according to different positions where each sum of housing price data belong to and different peripheral terms, shows the space distribution figure of the analysis result with GIS according to the parameter of neighborhood effects of urban park impact on hosing price.
