本研究計畫的目的有二：一、估算變電所設置的外部成本；二、比較分析傳統計量經濟模式與機器學習模式間的差異。本研究將利用特徵價格分析做為研究模式，透過收集房屋市場交易資料，並利用地理資訊系統建構房屋交易標地與變電所距離變數，探討分析變電所設置對於房屋價格的影響。在估計方法上，本研究除了採用傳統的一般最小平方法估計外，並將使用機器學習分析模式，特別是支持向量機方法，進而比較預測價格與距離間的關係。本研究的主題與研究方式，在國內過去的經濟研究上則有所欠缺，本研究成果不但具學術意義，研究成果並可以供做制定環境與能源政策的參考。 ;The purpose of this study has two goals: (1) to estimating the social costs of the establishment of power substation. (2) to comparing the results obtained between the traditional econometric methods and the machine learning methods. By adopting the hedonic pricing method, our research will collect the transaction data from the housing market on the cities of Taipei, New Taipei, Taoyuan, Taichung, Tainan and Kaohsiung. Furthermore, our work will use the geographical information system (GIS) to construct the distance variable for the measurement between the power substation and the housing location. In addition, the research work will use the machine learning method, in particular, the support vector machine beside the traditional ordinary least squares estimation. The research will examine and compare the results obtained from the two estimation methods. Due to the few studies related to estimate the social costs of the establishment of power substation in Taiwan, our work has not only play the role for the academic research but also provide the quantify results for the design the suitable environmental and energy policy.
