F检验--两个正态总体方差检验
作者:互联网
方差比的区间估计
假设:样本
X
1
,
.
.
.
,
X
n
1
X_1,...,X_{n_1}
X1,...,Xn1来自总体
X
∼
N
(
μ
1
,
σ
1
2
)
X\sim N(\mu_1,\sigma_1^2)
X∼N(μ1,σ12),样本
Y
1
,
.
.
.
,
Y
n
2
Y_1,...,Y_{n_2}
Y1,...,Yn2来自总体
Y
∼
N
(
μ
1
,
σ
2
2
)
Y\sim N(\mu_1,\sigma_2^2)
Y∼N(μ1,σ22)。样本均值与方差分别标记为,
X
ˉ
=
1
n
1
∑
i
=
1
n
1
X
i
,
Y
ˉ
=
1
n
2
∑
i
=
1
n
2
Y
i
\bar{X}=\frac{1}{n_1}\sum_{i=1}^{n_1}X_i,\qquad \bar{Y}=\frac{1}{n_2}\sum_{i=1}^{n_2}Y_i
Xˉ=n11i=1∑n1Xi,Yˉ=n21i=1∑n2Yi
和
S
1
2
=
1
n
1
−
1
∑
i
=
1
n
1
(
X
i
−
X
ˉ
)
2
;
S
2
2
=
1
n
2
−
1
∑
i
=
1
n
2
(
Y
i
−
Y
ˉ
)
2
S_1^2=\frac{1}{n_1-1}\sum_{i=1}^{n_1}\left( X_i-\bar{X} \right)^2;\qquad S_2^2=\frac{1}{n_2-1}\sum_{i=1}^{n_2}\left( Y_i-\bar{Y} \right)^2
S12=n1−11i=1∑n1(Xi−Xˉ)2;S22=n2−11i=1∑n2(Yi−Yˉ)2
当两样本独立时,
S
1
2
/
σ
1
2
S
2
2
/
σ
2
2
∼
F
(
n
1
−
1
,
n
2
−
2
)
\frac{S_1^2/\sigma_1^2}{S_2^2/\sigma_2^2}\sim F(n_1-1,n_2-2)
S22/σ22S12/σ12∼F(n1−1,n2−2)
则
σ
1
2
/
σ
2
2
\sigma_1^2/\sigma_2^2
σ12/σ22的置信水平
1
−
a
1-a
1−a的置信区间为,
[
S
1
2
/
S
2
2
F
a
/
2
(
n
1
−
1
,
n
2
−
1
)
,
S
1
2
/
S
2
2
F
1
−
a
/
2
(
n
1
−
1
,
n
2
−
1
)
]
\left[ \frac{S_1^2/S_2^2}{F_{a/2}(n_1-1,n_2-1)},\quad\frac{S_1^2/S_2^2}{F_{1-a/2}(n_1-1,n_2-1)} \right]
[Fa/2(n1−1,n2−1)S12/S22,F1−a/2(n1−1,n2−1)S12/S22]
方差比假设检验
假设零假设与备择假设为,
H
0
:
σ
1
2
/
σ
2
2
=
1
,
H
1
:
σ
1
2
/
σ
2
2
≠
1
H_0:\sigma_1^2/\sigma_2^2=1,\quad H_1: \sigma_1^2/\sigma_2^2\neq1
H0:σ12/σ22=1,H1:σ12/σ22=1
当零假设为真时,即
σ
1
=
σ
2
\sigma_1=\sigma_2
σ1=σ2,有
S
1
2
/
σ
1
2
S
2
2
/
σ
2
2
=
S
1
2
S
2
2
∼
F
(
n
1
−
1
,
n
2
−
1
)
\frac{S_1^2/\sigma_1^2}{S_2^2/\sigma_2^2}=\frac{S_1^2}{S_2^2}\sim F(n_1-1,n_2-1)
S22/σ22S12/σ12=S22S12∼F(n1−1,n2−1)
记统计量
F
=
S
1
2
S
2
2
F=\frac{S_1^2}{S_2^2}
F=S22S12,则有
{
F
≥
F
a
/
2
(
n
1
−
1
,
n
2
−
1
)
,
拒
绝
原
假
设
H
0
;
F
≤
F
1
−
a
/
2
(
n
1
−
1
,
n
2
−
1
)
,
拒
绝
原
假
设
H
0
.
\left\{\begin{array}{ll} F\geq F_{a/2}(n_1-1,n_2-1),& 拒绝原假设H_0;\\ F\leq F_{1-a/2}(n_1-1,n_2-1),& 拒绝原假设H_0. \end{array} \right.
{F≥Fa/2(n1−1,n2−1),F≤F1−a/2(n1−1,n2−1),拒绝原假设H0;拒绝原假设H0.
拒绝域为,
(
0
,
F
1
−
a
/
2
(
n
1
−
1
,
n
2
−
1
)
)
,
(
F
a
/
2
(
n
1
−
1
,
n
2
−
1
)
,
+
∞
)
(0,F_{1-a/2}(n_1-1,n_2-1)),\quad (F_{a/2}(n_1-1,n_2-1),+\infty)
(0,F1−a/2(n1−1,n2−1)),(Fa/2(n1−1,n2−1),+∞)
该检验是通过 F F F统计量的计算来实现的,也称之为 F F F检验。
标签:frac,--,S12,正态,检验,S22,n1,n2,sigma 来源: https://blog.csdn.net/lqj_614_19/article/details/123589943