不同最小平方和法分析花卉資料

Abstract

最小平方和法是最實用的迴歸方法，而近似無相關迴歸、兩階段最小平方和法和三階段最小平方和法則是最小平方和法的延伸。本論文歸納出迴歸方法的判別流程，並利用判別流程為不同的花卉模型做迴歸方法選擇。首先，為節慶花卉康乃馨找出最適合的迴歸估計方法為兩階段最小平方和法，並發現將節慶量化分級可做為康乃馨的量測變數，此研究可提供花卉運銷決策者將節慶量化做為康乃馨價格預測來源。接著，為台灣十大花卉迴歸模型找出最適合的迴歸估計方法為三階段最小平方和法，並證實花卉的每日交易資料存在互相影響，此研究幫助花卉業者掌握隔天台灣十大花卉各別的拍賣均價落點，也提供拍賣員做為每日花卉拍賣開價的依據。

The Least Squares Method is a common regression method. Seemingly Unrelated Regression, Two-Stage Least Squares Method, and Three-Stage Least Squares Method are extensions of The Least Squares Method. First, the work proposes a judge flow of those regression methods. Using the flow chart, user can choose methods for different models of flower species. Second, the work separately determines the regression methods for models of Carnation and top ten flowers in Taiwan. At last, the specified regression methods estimate the auction price for the flowers and Carnation in special festival. The information helps predicting the revenue for farmers and determining the beginning bid price for bidders.