Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/53748
標題: 具類別與連續變數之無參數分量迴歸-估計、檢定與應用
Nonparametric Quantile Regressions with Categorical and Continuous Variables: Estimation, Inference, and Applications
作者: 陳美源
關鍵字: 應用研究;categorical variable;經濟學;類別變數;無參數迴歸;分量迴歸;顯著性檢定;平滑參數;nonparametric regression;quantile regression;significance test;bandwidth selection
摘要: 
本研究計畫擬以三年的研究期間,完整而深入探討同時具有類別(categorical)及連續(continuous)變數無參數份量迴歸函數(nonparametric quantile regression function)之估計方法、顯著性檢定方法及其實證應用的研究;本研究計畫將擴展Racine andLi (2005)所提出對同時具有類別與連續變數的無參數迴歸函數的估計方法與平滑參數(smoothing parameter)決定方法,應用於無參數份量迴歸函數的估計上,本研究計畫將分別探討類別變數為解釋變數與應變數的迴歸函數;再者,本研究將應用Racine(1997)及Ait-Sahalia et al.(2001)所提出在無參數迴歸函數的顯著性檢定方法於同時具有類別與連續變數的無參數份量迴歸函數的檢定上;最後,根據先驗建立的估計方法與顯著性檢定方法的理論基礎,本研究計畫將對台灣房屋仲介商對銷售期間(time-on-market)的影響效果、台灣某家銀行發行現金卡違約風險機率的衡量、及台灣金融機構信用評等的建立等問題進行實證研究。

In this research, estimations, significance tests, and empiricalapplications of nonparametric quantile regressions function with bothcategorical and continuous variables are comprehensively studied.Typically, the estimation method and the data-driven bandwidth selectionsuggested by Racine and Li~(2005) for nonparametric regressions areextended to nonparametric quantile regression function with categoricaland continuous variables. Discrete variable(s) as a dependent variable andas an independent variable are the regression functions under investigated.Besides, the significance tests proposed by Racine(1997)及Ait-Sahalia etal.(2001) for nonparametric regression function are also extended to thenonparametric quantile regression function with both categorical andcontinuous variables. Empirical applications for studying the effect ofa broker on the time-on-market of a house selling, the default rate of cashcards, and the credit scoring for Taiwan』s financial institutes are goingto be investigated through the nonparametric quantile regressions.
URI: http://hdl.handle.net/11455/53748
其他識別: NSC96-2415-H005-001
Appears in Collections:財務金融學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.