本研究欲探討從消費者購買汽車決定重要屬性的大數據資料中，利用Relief-F 算法篩選出關鍵的重要汽車屬性；利用此汽車屬性製作成問卷加以發放，再針對問卷回收後執行統計逐步迴歸分析，得到符合國內消費者真正對購買車的重要屬性。研究結果從二資料庫共有34 個特徵，經過Relief-F 特徵過濾與逐步迴歸法篩選後，最後萃取出9 個消費者心中認為最重要的汽車屬性。即可再進一步利用ANOVA 檢定出顧客基本的特性與購買的重要因素之關係，即可分析出汽車市場的區隔。市場區隔有了清楚的定義，各家車廠就能善用資源針對自己的市場去分析，同時也能定義出直接競爭者、潛在競爭者及替代品競爭者，形成SWOT 分析的基礎。

In this study, we use Relief-F algorithm to filter the main attributes from the big data of consumers purchase automobile, and then use this key attributes to make into questionnaires. Then analysis the results by stepwise regression method, we got the really the 9 attributes of the consumers who cared in automobile purchase. Finally, we use ANOVA method to determine the basic characteristics of the customer and the purchase of the relationship between the important attributes, we can analysis the segment of the automobile market. Automobile marketing segments have a clear definition, each automobile-manufacturers can focus all resources to develop their marketing, to beat potential competitors.