【概率论基础进阶】参数估计-点估计

定义：用样本$X_1,X_2,\cdots ,X_n$构造的统计量$\hat{\theta }(X_1,X_2,\cdots ,X_n)$来估计参数$\theta$称为点估计，统计量$\hat{\theta }(X_1,X_2,\cdots ,X_n)$称为估计量

估计量是随机变量，它所取得的观测值$\hat{\theta }(x_1,x_2,\cdots ,x_n)$称为估计量，有时将$\theta$的估计量和估计值统称为$\theta$的估计

定义：设$\hat{\theta }$是$\theta$的估计量，如果$E(\hat{\theta })$，则称$\hat{\theta }=\hat{\theta }(X_1,X_2,\cdots ,X_n)$是未知参数$\theta$的无偏估计量

例1：设总体$X$的$E(X)=\mu ,D(X)={\sigma }^2$，从来自总体$X$的样本$X_1,X_2,\cdots ,X_n$得样本均值$\bar{X}=\frac{1}{n}\sum _{i=1}^n{X}_i$和样本方差$S^2=\frac{1}{n-1}\sum _{i=1}^n(X_i-\bar{X}{)}^2$，试证$S^2$是${\sigma }^2$的无偏估计

$$\begin{array}{rl}E(S^2)& =\frac{1}{n-1}E\left\{\sum _{i=1}^n[(X_i-\mu )-(\bar{X}-\mu ){]}^2\right\}\\ & =\frac{1}{n-1}E\left\{\sum _{i=1}^n[(X_i-\mu {)}^2-2(X_i-\mu )(\bar{X}-\mu )+(\bar{X}-\mu {)}^2]\right\}\\ & =\frac{1}{n-1}E\left[\sum _{i=1}^n(X_i-\mu {)}^2-n(\bar{X}-\mu {)}^2\right]\\ & =\frac{1}{n-1}\left[\sum _{i=1}^{n}E(X_i-\mu {)}^2-nE(\bar{X}-\mu {)}^2\right]\\ & =\frac{1}{n-1}[n{\sigma }^2-nD(\bar{X})]\\ & ={\sigma }^2\end{array}$$

之前做过用$\sum _{i=1}^n(X_i-\bar{X}{)}^2=\sum _{i=1}^n{X}_i^2-n{\bar{X}}^2$做的，这里不再赘述
$$\begin{array}{rl}\sum _{i=1}^n(X_i-\bar{X}{)}^2& =\sum _{i=1}^n{X}_i^2-n{\bar{X}}^2\\ \sum _{i=1}^n(X_i-\bar{X}{)}^2& =\sum _{i=1}^n(X_i-\mu {)}^2-n(\bar{X}-\mu {)}^2\end{array}$$
这两个公式都很重要

定义：设$\widehat{{\theta }_1}$和$\widehat{{\theta }_2}$都是$\theta$的无偏估计量，且$D(\widehat{{\theta }_1})\le D(\widehat{{\theta }_2})$，则称$\widehat{{\theta }_1}$比$\widehat{{\theta }_2}$更有效，或$\widehat{{\theta }_1}$比$\widehat{{\theta }_2}$更有效估计量

例2：设总体的数学期望和方差分别为$\mu$和${\sigma }^2$，$X_1,X_2$是来自总体$X$的样本。记$\tilde{X}=(1-a)X_1+a{X}_2$，确定$a$，使$D(\tilde{X})$最小

$$\begin{array}{rl}E(\tilde{X})& =(1-a)E(X_1)+aE(X_2)=\mu \\ D(\tilde{X})& =(1-a{)}^{2}D(X_1)+a^{2}D(X_2)\\ & =[(1-a{)}^2+a^2]{\sigma }^2\end{array}$$
显然当$\begin{array}{r}a=\frac{1}{2}\end{array}$时$D(\tilde{X})$最小

可以证明如果样本为$X_1,X_2,\cdots ,X_n$时，$\mu$的估计$\begin{array}{r}\bar{X}=\frac{1}{n}\sum _{i=1}^n{X}_i\end{array}$是形如
$$\tilde{X}=\sum _{i=1}^n{a}_i{X}_i(其中\sum _{i=1}^n{a}_i=1)$$
的估计中最有效的

定义：设$\tilde{\theta }(X_1,X_2,\cdots ,X_n)$是$\theta$的估计，如果$\tilde{\theta }$依概率收敛于$\theta$，则称$\tilde{\theta }(X_1,X_2,\cdots ,X_n)$为$\theta$的一致估计量

例3：设总体$X$的概率密度为
$$f(x;\theta )=\{\begin{array}{rlr}& \frac{2x}{3\theta }& \theta <x<2\theta \\ & 0& 其他\end{array}$$
其中$\theta$是未知参数，$X_1,X_2,\cdots ,X_n$为来自总体的简单随机样本，已知$E(c\sum _{i=1}^n{X}_i^2)={\theta }^2$，则$c=()$

$$\begin{array}{rl}E(c\sum _{i=1}^n{X}_i^2)& =c\sum _{i=1}^{n}E(X_i^2)\\ & =cnE(X^2)\\ & =cn{\int }_{\theta }^{2\theta }{x}^{2}f(x;\theta )dx\\ & =cn{\int }_{\theta }^{2\theta }\frac{2x^3}{3{\theta }^2}dx\\ & =cn\cdot \frac{5}{2}{\theta }^2={\theta }^2\Rightarrow c=\frac{2}{5n}\end{array}$$

例4：设总体$X$的概率密度为
$$f(x)=\{\begin{array}{rlr}& 2e^{-x(x-\theta )}& x>\theta \\ & 0& x\le \theta \end{array}$$
其中$\theta >0$是未知参数，从总体$X$中抽取简单随机样本$X_1,X_2,\cdots ,X_n$，记 θ ^ = min ⁡ { X 1 , X 2 , ⋯   , X n } \hat{\theta}=\min \left\{X_{1},X_{2},\cdots,X_{n}\right\} θ^=min{X1​,X2​,⋯,Xn​}

• 求总体$X$的分布函数$F(x)$

F ( x ) = P { X ≤ x } = ∫ − ∞ x f ( x ) d x = { ∫ 0 x 2 e − 2 ( x − θ ) d x = 1 − e − 2 ( x − θ ) x > θ 0 x ≤ θ \begin{aligned} F(x)&=P \left\{X \leq x\right\}\\ &=\int_{-\infty}^{x}f(x)dx\\ &=\left\{\begin{aligned}&\int_{0}^{x}2e^{-2(x-\theta)}dx=1-e^{-2(x-\theta)}&x>\theta\\&0&x \leq \theta\end{aligned}\right. \end{aligned} F(x)​=P{X≤x}=∫−∞x​f(x)dx=⎩ ⎨ ⎧​​∫0x​2e−2(x−θ)dx=1−e−2(x−θ)0​x>θx≤θ​​

• 求统计量$\hat{\theta }$的分布函数$F_{\hat{\theta }}(x)$

F θ ^ ( X ) = P { θ ^ ≤ x } = P { min ⁡ ( X 1 , X 2 , ⋯   , X n ) ≤ x } = 1 − P { min ⁡ ( X 1 , X 2 , ⋯   , X n ) > x } = 1 − P { X 1 > x , X 2 > x , ⋯   , X n > x } = 1 − P { X 1 > x } P { X 2 > x } ⋯ P { X n > x } = 1 − [ 1 − F ( x ) ] n = { 1 − e 2 n ( x − θ ) x > θ 0 x ≤ θ \begin{aligned} F_{\hat{\theta}}(X)&=P \left\{\hat{\theta}\leq x\right\}\\ &=P \left\{\min (X_{1},X_{2},\cdots,X_{n})\leq x\right\}\\ &=1-P \left\{\min (X_{1},X_{2},\cdots,X_{n})>x\right\}\\ &=1-P \left\{X_{1}>x,X_{2}>x,\cdots ,X_{n}>x\right\}\\ &=1-P \left\{X_{1}>x\right\}P \left\{X_{2}>x\right\}\cdots P \left\{X_{n}>x\right\}\\ &=1-[1-F(x)]^{n}\\ &=\left\{\begin{aligned}&1-e^{2n(x-\theta)}&x>\theta\\&0&x \leq \theta\end{aligned}\right. \end{aligned} Fθ^​(X)​=P{θ^≤x}=P{min(X1​,X2​,⋯,Xn​)≤x}=1−P{min(X1​,X2​,⋯,Xn​)>x}=1−P{X1​>x,X2​>x,⋯,Xn​>x}=1−P{X1​>x}P{X2​>x}⋯P{Xn​>x}=1−[1−F(x)]n={​1−e2n(x−θ)0​x>θx≤θ​​

• 如果用$\hat{\theta }$作为$\theta$的估计量，讨论它是否具有性质$E\hat{\theta }=\theta$

$$\begin{array}{rl}{f}_{\hat{\theta }}(x)& =F_{\hat{\theta }}^{\prime }(x)=\{\begin{array}{rlr}& 2n{e}^{-2n(x-\theta )}& x>\theta \\ & 0& x\le \theta \end{array}\\ E\hat{\theta }& ={\int }_{-\infty }^{+\infty }x{f}_{\hat{\theta }}(x)dx\\ & ={\int }_0^{+\infty }x\cdot 2n{e}^{-2n(x-\theta )}dx\\ & =\theta +\frac{1}{2\theta }\ne \theta \end{array}$$
所以不具有性质$E\hat{\theta }=\theta$

CSDN话题挑战赛第2期
参赛话题：学习笔记