在傳統實證分析中，最常被採用的次市場定義，是依先驗主觀認定的方式加以區隔，如以行政地理疆界或住宅特質來界定住宅次市場的範圍。雖然以先驗的主觀看法來定義次市場是蠻重要的，但是，如果以住宅代替性高低來定義次市場範圍模式，則實際住宅次市場可能有其他更合理定義方式。有鑑於此，本文以因素分析法，以大台北都會區爲例，探討住宅次市場定義之合理性。實證結果發現，依據2004年內政部地政司「房地產交易價格簡訊」資料庫，可將大台北地區住宅市場篩選出三個共同因素，分別爲住宅結構因素、總價坪數因素、與單價街寬因素，而且利用這三個共同因素，可將大台北都會區市場區隔爲三個次市場，且這三個次市場在統計上均具有顯著的差異。特別是在住宅結構因素上，與傳統行政區域劃分方式相較，新的統計方法的確能降低區域內的差異；亦即當此種方式至新定義次市場，可以使房價得到更好的估計結果。若我們再把台北市與台北縣的資料，利用因素分析法的個別市場方式，區分成三個次市場，來估計個別的房價，結果我們發現平均絕對估計誤差變小與命中率上升。換言之，新的統計方法不但可以當成另外一種次市場的定義，也可以當成傳統次市場的補充方法，此種互補的結果可以使我們對房價的估計準確度提高。 Previously, submarkets were defined based on the geographical boundaries or physical characteristics of the dwellings using a prior decision or predefined. Recently, there has been generally agreement that housing submarkets should be defined statistically to optimize the accuracy of hedonic predictions. This study first applies multivariate methods to define housing submarkets, and then explores the alternative definitions of submarkets on prediction accuracy. The empirical analysis employs an actual transaction data set from Taipei Metropolitan in 2004. This study found that the housing market of the Taipei Metropolitan area can be divided into three submarkets, with three common latent factors: housing construction, total housing price and floor space, and housing unit price and street width, respectively. Statistics differ significantly among the three submarkets. Compared with the traditional method by an administrative area, the inter-area difference of the housing construction factor is reduced. On the other hand, this study applies factor analysis using data for Taipei City and Taipei County, respectively. This study found that, when the separated data set was divided for both Taipei City and Taipei County, the hit rates were considerably higher while the MAPEs were significantly lower. The empirical results demonstrate that the three submarkets among city (or county) could be a good compliment for improving the prediction accuracy.