【背景】
在现实生活中,我们做具体的数理统计时,常已明确知道了总体X的分布,但却不能准确地得知分布的参数,也就是参数未知,因此,在实际应用中我们需要对参数做出估计。
以下介绍几种常用的参数估计方法。
【点估计法】
<1>矩估计法:
[知识扩展]
\(\begin{align}&E(X) \rightarrow 总体的一阶原点矩 \\&E(X^2) \rightarrow总体的二阶原点矩 \\ &\bar{X} = \frac{1}{n} \sum_{i=1}^{n} x_i = A_1 \rightarrow 样本的一阶原点矩 \\&\frac{1}{n} \sum_{n}^{i=1} x_i ^2 = A_2 \rightarrow 样本的二阶原点矩 \\& \frac{1}{n} \sum_{n}^{i=1}x_i \rightarrow E(X) \\& \frac{1}{n} \sum_{n}^{i=1}x_i^2 \rightarrow E(X^2)\end{align}\)
\(\begin{align}&case1:未知参数只含\theta时:\\&①求出E(X) \\&②令E(X)= \bar{X} \\&③得出\widehat{\theta}的表达式\\ \\&case2:总体含有 \theta_1 \theta_2 \\&①求出E(X) 和 E(X^2) \\&②令 \begin{cases} E(X) = \bar{X} \\ E(X^2) = A_2 \end{cases} \Rightarrow \begin{cases}\widehat{\theta_1}= \\ \widehat{\theta_2} = \end{cases} (关于\widehat{\theta_1}, \widehat{\theta_2}的关系式)\end{align}\)
<2>最大似然估计法
\(\begin{align}&cases1.对于离散分布已知分布律。\\&①X \rightarrow (X_1,X_2,\cdots,X_n) \Rightarrow (x_1,x_2,\cdots,x_n) \\&②求L(\theta) = P\{x_1 = k_1\} \cdots P\{x_n = k_n\} =P\{X_1=k_1\} \cdots P\{X = k_n\} \\&(利用对数运算规律,将乘除法转换为加减法,便于求导运算) \\&③令\frac{d}{d\theta} (\ln L(\theta)) = \cdots = 0 \\&④得出\widehat{\theta} 的表达式 \\ \\&case2:总体X服从于f(x;\theta)\\&① X \Rightarrow (X_1,\cdots,X_n) \Rightarrow (x_1,\cdots,x_n) \\&②L(\theta) = f(x_1;\theta)f(x_2;\theta) \cdots f(x_n;\theta) \\ &(利用对数运算规律,将乘除法转换为加减法,便于求导运算) \\&③令\frac{d}{d\theta} (\ln L(\theta)) = \cdots=0 \\&④得出\widehat{\theta}的表达式\end{align}\)
【估计量的评价标准】
\(\begin{align}&①无偏性-X总体(含未知参数\theta) \\&(X_1,X_2,\cdots,X_n)为样本\\&若\widehat{\theta} = \varphi(X_1,X_2,\cdots,X_n)为\theta的估计量\\&若E(\widehat{\theta}) = \theta,则称\widehat{\theta} = \varphi(X_1,X_2,\cdots,X_n)为\theta的无偏估计量\\ \\&②有效性-设\widehat{\theta} = \varphi_1(X_1,X_2,\cdots,X_n) \\&\widehat{\theta} = \varphi_2(X_1,X_2,\cdots,X_n)皆为\theta的无偏估计量\\&即E(\varphi_1(X_1,X_2,\cdots,X_n)) = E(\varphi_2(X_1,X_2,\cdots,X_n)) \\&若D(\varphi_1(X_1,X_2,\cdots,X_n)) < D(\varphi_2(X_1,X_2,\cdots,X_n)) \\&称\widehat{\theta} = \varphi_1(X_1,X_2,\cdots,X_n)是比\widehat{\theta} = \varphi_2(X_1,X_2,\cdots,X_n)更有效的估计量\end{align} \)
【区间估计】
\(\begin{align}&X服从于N(u,\sigma ^2) \Rightarrow (X_1,X_2,\cdots,X_n) \\&①对u估计(置信估计1-\alpha) \\&case1:\sigma已知\\&1.U=\frac{\bar{X} – \mu}{\frac{\sigma}{\sqrt{n}}} 服从于N(0,1) \\&2.查表\pm Z_{\frac{\alpha}{2}} \\&3.P\{-Z_{\frac{\alpha}{2}} < \frac{\bar{X} – \mu}{\frac{\sigma}{\sqrt{n}}}\} = 1 -\alpha \\& \mu的置信度为1-\alpha的置信区间为:\\& (\bar{X} – \frac{\sigma}{\sqrt{n}}Z_{\frac{\alpha}{2}},\bar{X}+\frac{\sigma}{\sqrt{n}}Z_{\frac{\alpha}{2}}) \\ \\&case2: \sigma未知 \\&1.t = \frac{\bar{X} – \mu}{\frac{s}{\sqrt{n}}} 服从于t(n-1) \\&2.查表\pm t_{\frac{\alpha}{2}}(n-1) \\&3.P\{-t_{\frac{\alpha}{2}}(n-1) < \frac{\bar{X} -\mu}{\frac{s}{\sqrt{n}}} < t_{\frac{\alpha}{2}}(n-1)\} = 1 – \alpha \\& \mu的置信度为1-\alpha的置信区间为\\& (\bar{X} – \frac{s}{\sqrt{n}}t_{\frac{\alpha}{2}}(n-1), \bar{X} + \frac{s}{\sqrt{n}}t_{\frac{\alpha}{2}}(n-1)) \\ \\&②对\sigma ^2估计\\&case1:\mu已知\\&1.\frac{1}{\sigma^2}\sum_{n}^{i=1}(X_i – \mu)^2 服从于x^2(n) \\&2.查表x_{1-\frac{\alpha}{2}}^2(n)=? 和x_{\frac{\alpha}{2}}^2(n) =?\\&3.P\{x_{1-\frac{\alpha}{2}}^2(n)<\frac{1}{\sigma^2}\sum_{i=1}^{n} (x_i-\mu)^2 < x_{\frac{\alpha}{2}}^2(n)\} = 1 – \alpha \\& \sigma^2的置信区间为:(\frac{\sum_{i=1}^{n}(x_i – \mu)^2}{x_{\frac{\alpha}{2}}^2(n)},\frac{\sum_{i=1}^{n}(x_i-\mu)^2}{x_{1-\frac{\alpha}{2}}^2(n)}) \\ \\&case2:\mu未知 \\&1.\frac{(n-1)s^2}{\sigma^2}服从于x^2(n-1) \\&2.x_{1-\frac{\alpha}{2}^2}(n-1) =?和x_{\frac{\alpha}{2}^2}(n-1) =? \\&3.P\{x_{1 – \frac{\alpha}{2}^2}(n-1) < \frac{(n-1)s^2}{\sigma^2} < x_{\frac{\alpha}{2}^2}(n-1)\} \\&\sigma ^2的置信区间为\\&(\frac{(n-1)s^2}{x_{\frac{\alpha}{2}}(n-1)},\frac{(n-1)s^2}{x_{1-\frac{\alpha}{2}}(n-1)})\end{align}\)
文章导航