本研究蒐集台南市區某房仲業者2012年12月至2015年8月之實際交易資料共189筆,主旨在於分析影響台南市房地產價格因素。與過往研究不同者,本研究認為,鄰里特徵影響房價甚巨,如就學學區因明星學校及其升學率、學校教學策略等因素,使之成為部分家庭選擇居所之重要考量;又如住宅四周之環境與建設特徵影響居住民眾之生活品質與便利性,亦成為國人購屋的重要考量。 本研究除傳統房地產價格因素如:地坪、衛浴、屋況、樓層等戶棟特徵外,另加入環境變數與學區變數等鄰里特徵用以分析房地產價格,透過最小平方迴歸、分量迴歸、與Cubist迴歸樹方法進行統計分析比較,並建立台南地區房價預測模型。結果顯示,Cubist迴歸樹模型整體準確度表現為最佳。影像房地產價格變數的重要程度依序為建坪數、嫌惡設施距離、屋齡、百貨距離、衛浴數、地坪數、廟宇距離,其中嫌惡設施距離、百貨距離、廟宇距離為鄰里環境變數。
We collected 189 realty data in Tainan area from December 2012 to August 2015. The purpose is to analyze the factors affecting the price of real estate in Tainan City. The differences from the past studies, is to consider the roles of neighborhood characteristics to the price of houses. Because of famous schools, the school enrollment rate, school teaching strategies and other factors, make an important parts of the family's choice of residence. And the environment and construction characteristics affect the quality and convenience of living, also become an important consideration for homebuyers. In this study, except traditional real estate price factors, such as land pings, bathrooms, house conditions, floors, and other buildings features, we also added environment variables like school district variables and other neighborhoods characteristics to analyze the real estate prices. By using ordinary least squares regression, quantile regression, and Cubist regression tree methods to do statistical analyzing. In the results, Cubist regression trees model has best accuracy overall. The variables which affect real estate price arrange in order of importance are total pings, distance from NIMBY facilities, house age, distance from department store, the number of bathrooms, land pings, distance from temple. Distance from temple, distance from NIMBY facilities, and the distance from department store are neighborhoods environment variables.