透過您的圖書館登入
IP:3.19.56.114
  • 學位論文

分布函數之反函數之核估計的模擬研究

A Simulation Study for Kernel Estimator of Inverse Distribution Function

指導教授 : 許玉生
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


分布函數為機率上重要的分析工具,其重要性不亞於機率密度函數及特徵函數。在統計上分布函數也有很多應用,令$F$表一分布函數,則$F^{-1}$可用於隨機變數之模擬及穩定型分布(stable distribution)之冪數(exponent)的估計。通常分布函數是未知的,必需用樣本估計。分布函數未知時,常用之分布函數的估計式為經驗分布(empirical distribution function)。本文之目的為研究$F^{-1}$的估計,但上述經驗分布卻因其反函數不存在,故不能直接運用。本文提出$F^{-1}(y)$之核估計式$widehat{F}^{-1}(y)$,因此式之機率性質非常複雜,故本文將以電腦模擬方式研究$widehat{F}^{-1}(y)$之漸近一致性(asymptotic consistency)及漸近常態性(asymptotic normality)。

並列摘要


The inverse function of a distribution function has many applications in statistics. In practice, the inverse function is unknown and has to be estimated. The purpose of this paper is to discuss a kernel estimator $widehat{F}^{-1}(y)$ of the inverse function $F^{-1}(y)$ of a distribution function $F(x)$. Since the theoretical property of $widehat{F}^{-1}(y)$ is extremely complicated, we will investigate the asymptotic consistency and asymptotic normality of $widehat{F}^{-1}(y)$ via computer simulations.

參考文獻


[1] 周宗翰(2007). 單峰穩定型分布之冪數的經驗分布及核密度函數估計法。中央大學數學研究所碩士論文。
[2] Alexander,K.S.(1984). Probability inequalities for empirical processes and a law of the iterated logarithm. Ann. probab.12,1041-1067.
[3] Bolthausen,E.(1978). Weak convergence of an empirical process indexed by the closed convex subsets of $I^2$. Z. Wahrsch. Verw. Gebiete,43,173-181.
[6] Dahlhaus,R.(1988). Empirical spectral processes and their application to time series analysis. Stochastic Processes. Appl. ,30,69-83.
[7] Dudlry,R.M.(1978). Central limit theorems for empirical measures. Probab.6,899-929.

延伸閱讀