概率统计D 07.05 单侧置信限

$\color{blue}{\S 7.5 单侧置信限}$

$对于给定值\alpha(0 < \alpha < 1),若由样本(X_1, X_2, \cdots, X_n)\\ 确定的统计量\theta_1 = \theta_1(X_1, X_2, \cdots, X_n)满足\\ P\lbrace \theta > \theta_1 \rbrace = 1 - \alpha, \\ 则称随机区间(\theta_1, +\infty)为参数\theta的置信度为1 - \alpha的\\ 单侧置信区间,\theta_1称为置信度为1 - \alpha的单侧置信下限.$
$若统计量\theta_2 = \theta_2(X_1, X_2, \cdots, X_n)满足\\ P\lbrace \theta < \theta_2 \rbrace = 1 - \alpha, \\ 则称随机区间(-\infty, \theta_2)为参数\theta的置信度为1 - \alpha的\\ 单侧置信区间,\theta_2称为置信度为1 -\alpha的单侧置信上限.$

$例1.从某批灯泡中随机抽取5只作寿命试验,\\ 测得其寿命(单位:h)如下:$
$1050, 1100, 1120, 1250, 1280$
$设灯泡的寿命服从正态分布,试求均值的置\\ 信度为0.95的单侧置信区间(\theta_1, +\infty).$
$解:设灯泡寿命为X \sim N(\mu, \sigma^2),由于\\ t = \dfrac{\overline X - \mu}{S/\sqrt{n}} \sim t(n-1)$
$由1 - \alpha = 0.95查表得t_{\alpha}(n-1),使得$
$P\lbrace \dfrac{\overline X - \mu}{S/\sqrt{n}} < t_{\alpha}(n-1) \rbrace = 1 - \alpha,$
$即P\lbrace \mu > \overline X - \dfrac{S}{\sqrt{n}} t_{\alpha}(n-1) \rbrace = 1 - \alpha,$
$于是得\mu 的置信度为1 - \alpha的单侧置信区间为$
$\left( \overline X - \dfrac{S}{\sqrt{n}} t_{\alpha}(n-1), +\infty \right)$
$亦即\mu 的1 - \alpha置信下限为\overline X - \dfrac{S}{\sqrt{n}} t_{\alpha}(n-1)$

$本题中,n = 5, 1 - \alpha = 0.95, \alpha = 0.05, \overline x = 1160, \\ s = 99.75, t_{\alpha}(n-1) = t_{0.05}(4) = 2. 1318, 因此\mu的\\ 置信度为0.95的单侧置信区间为$
$\left( \overline x - \dfrac{s}{\sqrt{n}} t_{\alpha} (n-1), +\infty \right)$
$= \left( 1160 - \dfrac{99.78}{\sqrt{4}} \times 2.1318, +\infty \right)$
$= (1065, +\infty)$
$亦即\mu 的0.95置信下限为1065$