概率论重修笔记 11矩估计和极大似然估计.md

矩估计方法(课11)

• 求期望$u_1=E(x)=f(\theta )$
• 用$u_1表示\theta \theta =f^{-1}(u_1)$
• $表示\hat{\theta }=f^{-1}(\overline{X})$

极大似然估计方法

• 先构造$L(\theta )={\prod }_{i=1}^{n}f(x,\theta )$
• 再取对数$ln(L(\theta ))=ln({\prod }_{i=1}^{n}f(x,\theta ))$
• $$\begin{gathered} 求导,并令导数等于0 \\ \frac{d}{d\theta }=0 \end{gathered}$$
• 解出$\theta$
• 取对数公式$$\begin{gathered} ln{X}^A=AlnX \\ lnX\ast Y=lnX+lnY \end{gathered}$$

---

分布律如下

$X$	$1$	$2$	$3$
$P$	${\theta }^2$	$2\theta (1-\theta )$	$(1-\theta {)}^2$

求他的矩估计量

1. 先求期望$$\begin{gathered} u_1=E(X)={\sum }_{i=1}^n{x}_i{P}_i \\ =1\ast {\theta }^2+2\ast 2\theta (1-\theta )+3\ast (1-\theta {)}^2 \\ =4\theta +3-6\theta \\ =3-2\theta \end{gathered}$$
2. 用$$\begin{gathered} u_1表示\theta \\ \theta =\frac{3-u_1}{2} \end{gathered}$$
3. 用$$\begin{gathered} \hat{\theta }和X表示=\hat{\theta }=\theta \\ \theta =\frac{3-\overline{X}}{2} \end{gathered}$$

---

设$X$的概率密度函数$f(x,\theta )=\{\begin{array}{ll}\theta x^{\theta -1}& 0<x<1\\ 0& 其他\end{array}$

其中$\theta >0$为未知参数,$X_1,X_2,X_3...X_n$为样本

1. 求$\theta$的矩估计
2. 求$\theta$的极大似然估计量

先构造$L(\theta )={\prod }_{i=1}^{n}f(x,\theta )$

$$\begin{gathered} L(\theta ) \\ ={\prod }_{i=1}^{n}f(x,\theta ) \\ =\theta x_1^{\theta -1}\theta x_2^{\theta -1}\theta x_3^{\theta -1}...\theta x_n^{\theta -1} \\ =[{\theta }^n]\ast {\theta }_1^{\theta -1}\ast {\theta }_2^{\theta -1}\ast {\theta }_3^{\theta -1}...\ast {\theta }_n^{\theta -1} \end{gathered}$$

然后取对数

$ln(L(\theta ))=ln[{\theta }^n]+ln[{\theta }_1^{\theta -1}]+ln[{\theta }_2^{\theta -1}]+ln[{\theta }_3^{\theta -1}]...+ln[{\theta }_n^{\theta -1}]$

$ln(L(\theta ))=nln[\theta ]+(\theta -1){\sum }_{i=1}^{n}ln({\theta }_i)$

再求导,令导数为0,并求出$\hat{\theta }$

$$\begin{gathered} \frac{d}{d\theta }==0 \\ =\frac{n}{ln\theta }+{\sum }_{i=1}^{n}ln({\theta }_i)==0 \\ \hat{\theta }=-\frac{n}{{\sum }_{i=1}nln{X}_i} \end{gathered}$$