| 摘要: | 特徵價格理論模型被廣泛用於估計住宅價格。因此，本研究應用特徵價格法探討臺中市東區、西區、南區、北區透天住宅房價，並透過文獻回顧選出足以反映住宅價格之屬性變數。
本文採用2008年透天住宅交易價格資料為依變數，透過複迴歸分析，採取常用之直線型、半對數型、逆半對數型及雙邊對數模式進行樣本分析，瞭解房價與各變數間之關連性，藉此討論影響住宅價格的重要因素。
本研究結果歸納如下：
1.在總價方面，直線型函數模式（R²adj＝0.923）相對於半對數（R²adj＝0.832）、逆半對數（R²adj＝0.763）及雙邊對數函數模式（R²adj＝0.858）有較佳的模式解釋能力。而逆半對數函數模式中，環境特徵被排除。
2.在單價方面，雙邊對數函數模式（R²adj＝0.669）相對於直線型（R²adj＝0.605）、半對數（R²adj＝0.590）及逆半對數函數模式（R²adj＝0.631）有較佳的模式解釋能力。
3.另一方面，建物面積、土地面積、屋齡、裡地、臨路寬、西區、鄰近學校距離與新市政中心（新CBD）距離等變數在兩個（總價與單價）模型中均為顯著，是影響透天住宅房價之重要因素。影響住宅總價與單價最顯著的特徵均為土地面積。
The Hedonic price theory has been extensively used to estimate housing prices. Therefore, this study aimed to apply the Hedonic price theory to estimate housing prices of town houses in East district, West district, South district, and North district of Taichung city. By reviewing the related researches, attributes could be selected which reflect the housing price.
This research uses the housing market prices in 2008 as dependent variables, then, through the multiple regression analyses taking the usual linear form, semi-log form, inverse semi-log and double-log form to analyze, and understand the relationship between the housing prices and the variables, in order to reveal the importance factors in house prices.
The results of the research can be concluded as followings：
1. In total price of house, the linear form (R²adj＝0.923) is better than semi-log form (R²adj＝0.832)、inverse semi-log form(R²adj＝0.763) and double-log form (R²adj＝0.858) in estimation. Environment variables are excluded in inverse semi-log form.
2. In the unit of housing price, the double-log form (R²adj＝0.669) is better than linear form (R²adj＝0.605)、semi-log form (R²adj＝0.590) and inverse semi-log form (R²adj＝0.631) in estimation.
3. In the other side, the attributes including floor space, the square footage, the housing age, the relation location to the street, the road width, in the west district or not, the distance to schools, and the distance to the new city hall center (New CBD) are significant factors to affect prices in both models. The most marked feature affects the total price and the unit price is the area of land. |
