2.18设 ( X , Y ) = ( ( X 1 , Y 1 ) , ⋯ , ( X n , Y n ) ) (X,Y)= ((X_{1},Y_{1}),\cdots,(X_{n},Y_{n})) (X,Y)=((X1,Y1),⋯,(Xn,Yn))为取自两维正态总体的简单随机样本。求总体均值 μ X , μ Y \mu_X,\mu_Y μX,μY,总体方差 σ X , σ Y \sigma_X,\sigma_Y σX,σY和总体相关系数 ρ X Y \rho_{XY} ρXY的最大似然估计(对于两维正态总体参数的最大似然估计,除了利用一般的求多维正态总体参数最大似然估计的方法外,也可将两维分布密度写为一维的边缘分布和条件分布的乘积,并利用密度的这一表示式求出各参数的最大似然估计)。
参考答案:
方法一:(多维正态总体参数)设 x 1 , ⋯ , x n x_1,\cdots,x_n x1,⋯,xn是对应 X 1 , ⋯ , X n X_1,\cdots,X_n X1,⋯,Xn的样本值, y 1 , ⋯ , y n y_1,\cdots,y_n y1,⋯,yn是对应 Y 1 , ⋯ , Y n Y_1,\cdots,Y_n Y1,⋯,Yn的样本值。
总体 Z = d N ( μ , Σ ) Z\stackrel{d}{=}N(\mu,\Sigma) Z=dN(μ,Σ)
其中 Z = ( X Y ) , μ = ( μ X μ Y ) , Z=\begin{pmatrix}X \\ Y\end{pmatrix},\mu=\begin{pmatrix}\mu_X\\\mu_Y\end{pmatrix}, Z=(XY),μ=(μXμY), Σ = ( σ X 2 ρ X Y σ X σ Y ρ X Y σ X σ Y σ Y 2 ) \Sigma=\begin{pmatrix}\sigma_X^2 & \rho_{XY}\sigma_X\sigma_Y\\ \rho_{XY}\sigma_X\sigma_Y&\sigma_Y^2\end{pmatrix} Σ=(σX2ρXYσXσYρXYσXσYσY2)
总体
Z
Z
Z的密度函数为
p
(
z
)
=
1
(
2
π
)
2
∣
Σ
∣
exp
{
−
1
2
(
z
−
μ
)
′
Σ
−
1
(
z
−
μ
)
}
p(z)=\frac{1}{\sqrt{(2\pi)^2|\Sigma|}}\exp\{-\frac{1}{2}(z-\mu)'\Sigma^{-1}(z-\mu)\}
p(z)=(2π)2∣Σ∣1exp{−21(z−μ)′Σ−1(z−μ)}
似然函数为
L ( μ , Σ ) = ∏ i = 1 n p ( z i ) = 1 ( 2 π ) n ∣ Σ ∣ n 2 exp { − 1 2 ∑ i = 1 n ( z i − μ ) ′ Σ − 1 ( z i − μ ) } \begin{array}{rl}&L(\mu,\Sigma)=\prod_{i=1}^{n}\limits p(z_i)\\&=\frac{1}{(2\pi)^n|\Sigma|^{\frac{n}{2}}}\exp\{-\frac{1}{2}\sum_{i=1}^{n}\limits(z_i-\mu)'\Sigma^{-1}(z_i-\mu)\}\end{array} L(μ,Σ)=i=1∏np(zi)=(2π)n∣Σ∣2n1exp{−21i=1∑n(zi−μ)′Σ−1(zi−μ)}
对数似然函数为
l ( μ , Σ ) = ln L ( μ , Σ ) = − n ln ( 2 π ) − n 2 ln ∣ Σ ∣ − 1 2 ∑ i = 1 n ( z i − μ ) ′ Σ − 1 ( z i − μ ) \begin{array}{rl}l(\mu,\Sigma)&=\ln L(\mu,\Sigma)\\&=-n\ln(2\pi)-\frac{n}{2}\ln|\Sigma|\\&-\frac{1}{2}\sum_{i=1}^{n}\limits(z_i-\mu)'\Sigma^{-1}(z_i-\mu)\end{array} l(μ,Σ)=lnL(μ,Σ)=−nln(2π)−2nln∣Σ∣−21i=1∑n(zi−μ)′Σ−1(zi−μ)
由 { d l ( μ , Σ ) d μ = 1 2 ∑ i = 1 n Σ − 1 ( z i − μ ) = 0 d l ( μ , Σ ) d Σ = − n 2 Σ − 1 + 1 2 ∑ i = 1 n ( z i − μ ) ( z i − μ ) ′ ( Σ − 1 ) 2 = 0 \left\{\begin{array}{rl} \frac{dl(\mu,\Sigma)}{d\mu}&=\frac{1}{2}\sum_{i=1}^{n}\limits\Sigma^{-1}(z_i-\mu)=0\\ \frac{dl(\mu,\Sigma)}{d\Sigma}&=-\frac{n}{2}\Sigma^{-1}\\&+\frac{1}{2}\sum_{i=1}^{n}\limits(z_i-\mu)(z_i-\mu)'(\Sigma^{-1})^2=0 \end{array} \right. ⎩⎪⎪⎪⎪⎨⎪⎪⎪⎪⎧dμdl(μ,Σ)dΣdl(μ,Σ)=21i=1∑nΣ−1(zi−μ)=0=−2nΣ−1+21i=1∑n(zi−μ)(zi−μ)′(Σ−1)2=0
得 { μ = 1 n ∑ i = 1 n z i = z ˉ Σ = 1 n ∑ i = 1 n ( z i − μ ) ( z i − μ ) ′ = 1 n ∑ i = 1 n ( z i − z ˉ ) ( z i − z ˉ ) ′ \left\{\begin{array}{rl} \mu&=\frac{1}{n}\sum_{i=1}^{n}\limits z_i=\bar{z}\\ \Sigma&=\frac{1}{n}\sum_{i=1}^{n}\limits(z_i-\mu)(z_i-\mu)'\\&=\frac{1}{n}\sum_{i=1}^{n}\limits(z_i-\bar{z})(z_i-\bar{z})' \end{array} \right. ⎩⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎧μΣ=n1i=1∑nzi=zˉ=n1i=1∑n(zi−μ)(zi−μ)′=n1i=1∑n(zi−zˉ)(zi−zˉ)′
故
μ
X
,
μ
Y
,
σ
X
,
σ
Y
,
ρ
X
Y
\mu_X,\mu_Y,\sigma_X,\sigma_Y,\rho_{XY}
μX,μY,σX,σY,ρXY的最大似然估计量为
μ
^
X
=
X
ˉ
,
μ
^
Y
=
Y
ˉ
,
\widehat{\mu}_X=\bar{X},\widehat{\mu}_Y=\bar{Y},
μ
X=Xˉ,μ
Y=Yˉ,
σ
^
X
=
1
n
∑
i
=
1
n
(
X
i
−
X
ˉ
)
2
,
\widehat{\sigma}_X=\sqrt{\frac{1}{n}\sum_{i=1}^{n}\limits(X_i-\bar{X})^2},
σ
X=n1i=1∑n(Xi−Xˉ)2,
σ
^
Y
=
1
n
∑
i
=
1
n
(
Y
i
−
Y
ˉ
)
2
,
\widehat{\sigma}_Y=\sqrt{\frac{1}{n}\sum_{i=1}^{n}\limits(Y_i-\bar{Y})^2},
σ
Y=n1i=1∑n(Yi−Yˉ)2,
ρ
^
X
Y
=
∑
i
=
1
n
(
X
i
−
X
ˉ
)
(
Y
i
−
Y
ˉ
)
∑
i
=
1
n
(
X
i
−
X
ˉ
)
2
∑
i
=
1
n
(
Y
i
−
Y
ˉ
)
2
.
\widehat{\rho}_{XY}=\frac{\sum_{i=1}^{n}\limits(X_i-\bar{X})(Y_i-\bar{Y})}{\sqrt{\sum_{i=1}^{n}\limits(X_i-\bar{X})^2}\sqrt{\sum_{i=1}^{n}\limits(Y_i-\bar{Y})^2}}.
ρ
XY=i=1∑n(Xi−Xˉ)2i=1∑n(Yi−Yˉ)2i=1∑n(Xi−Xˉ)(Yi−Yˉ).
方法二:(将两维分布密度写为一维的边缘分布和条件分布的乘积)设 x 1 , ⋯ , x n x_1,\cdots,x_n x1,⋯,xn是对应 X 1 , ⋯ , X n X_1,\cdots,X_n X1,⋯,Xn的样本值, y 1 , ⋯ , y n y_1,\cdots,y_n y1,⋯,yn是对应 Y 1 , ⋯ , Y n Y_1,\cdots,Y_n Y1,⋯,Yn的样本值。
总体
Z
=
d
N
(
μ
,
Σ
)
Z\stackrel{d}{=}N(\mu,\Sigma)
Z=dN(μ,Σ)
其中
Z
=
(
X
Y
)
,
μ
=
(
μ
X
μ
Y
)
,
Z=\begin{pmatrix}X \\ Y\end{pmatrix},\mu=\begin{pmatrix}\mu_X\\\mu_Y\end{pmatrix},
Z=(XY),μ=(μXμY),
Σ
=
(
σ
X
2
ρ
X
Y
σ
X
σ
Y
ρ
X
Y
σ
X
σ
Y
σ
Y
2
)
\Sigma=\begin{pmatrix}\sigma_X^2 & \rho_{XY}\sigma_X\sigma_Y\\ \rho_{XY}\sigma_X\sigma_Y&\sigma_Y^2\end{pmatrix}
Σ=(σX2ρXYσXσYρXYσXσYσY2)
总体 Z Z Z的密度函数为 p ( z ) = 1 ( 2 π ) 2 ∣ Σ ∣ exp { − 1 2 ( z − μ ) ′ Σ − 1 ( z − μ ) } p(z)=\frac{1}{\sqrt{(2\pi)^2|\Sigma|}}\exp\{-\frac{1}{2}(z-\mu)'\Sigma^{-1}(z-\mu)\} p(z)=(2π)2∣Σ∣1exp{−21(z−μ)′Σ−1(z−μ)}
X X X的边缘密度为 p ( x ) = ∫ p ( z ) d y = 1 2 π σ X exp { − ( x − μ X ) 2 2 σ X 2 } p(x)=\int p(z) dy=\frac{1}{\sqrt{2\pi}\sigma_X}\exp\{-\frac{(x-\mu_X)^2}{2\sigma_X^2}\} p(x)=∫p(z)dy=2πσX1exp{−2σX2(x−μX)2}
Y Y Y的边缘密度为 p ( y ) = ∫ p ( z ) d y = 1 2 π σ Y exp { − ( y − μ Y ) 2 2 σ Y 2 } p(y)=\int p(z) dy=\frac{1}{\sqrt{2\pi}\sigma_Y}\exp\{-\frac{(y-\mu_Y)^2}{2\sigma_Y^2}\} p(y)=∫p(z)dy=2πσY1exp{−2σY2(y−μY)2}
X
X
X在
Y
=
y
Y=y
Y=y下的条件分布
p
(
x
∣
Y
=
y
)
=
p
(
z
)
p
(
y
)
=
1
2
π
1
−
ρ
X
Y
2
σ
X
exp
{
−
1
2
(
1
−
ρ
X
Y
2
)
[
(
x
−
μ
X
)
2
σ
X
2
−
2
ρ
X
Y
(
x
−
μ
X
)
(
y
−
μ
Y
)
σ
X
σ
Y
+
(
y
−
μ
Y
)
2
σ
Y
2
]
+
(
y
−
μ
Y
)
2
2
σ
Y
2
}
\begin{array}{rl} p(x|Y=y)&=\frac{p(z)}{p(y)}\\&=\frac{1}{\sqrt{2\pi}\sqrt{1-\rho_{XY}^2}\sigma_X}\exp\{-\frac{1}{2(1-\rho_{XY}^2)}\\&[\frac{(x-\mu_X)^2}{\sigma_X^2}-\frac{2\rho_{XY}(x-\mu_X)(y-\mu_Y)}{\sigma_X\sigma_Y}\\&+\frac{(y-\mu_Y)^2}{\sigma_Y^2}]+\frac{(y-\mu_Y)^2}{2\sigma_Y^2}\} \end{array}
p(x∣Y=y)=p(y)p(z)=2π1−ρXY2σX1exp{−2(1−ρXY2)1[σX2(x−μX)2−σXσY2ρXY(x−μX)(y−μY)+σY2(y−μY)2]+2σY2(y−μY)2}
似然函数为
L
(
μ
X
,
Σ
X
)
=
∏
i
=
1
n
p
(
x
i
)
=
1
(
2
π
)
n
2
σ
X
n
exp
{
−
∑
i
=
1
n
(
x
i
−
μ
X
)
2
2
σ
X
2
}
L
(
μ
Y
,
Σ
Y
)
=
∏
i
=
1
n
p
(
y
i
)
=
1
(
2
π
)
n
2
σ
Y
n
exp
{
−
∑
i
=
1
n
(
y
i
−
μ
Y
)
2
2
σ
Y
2
}
L
(
ρ
X
Y
)
=
1
(
2
π
)
n
2
(
1
−
ρ
X
Y
2
)
n
2
σ
X
n
⋅
exp
{
−
1
2
(
1
−
ρ
X
Y
2
)
∑
i
=
1
n
[
(
x
i
−
μ
X
)
2
σ
X
2
−
2
ρ
X
Y
(
x
i
−
μ
X
)
(
y
i
−
μ
Y
)
σ
X
σ
Y
+
(
y
i
−
μ
Y
)
2
σ
Y
2
]
+
∑
i
=
1
n
(
y
i
−
μ
Y
)
2
2
σ
Y
2
}
\begin{array}{rl} L(\mu_X,\Sigma_X)&=\prod_{i=1}^{n}\limits p(x_i)\\&=\frac{1}{(2\pi)^{\frac{n}{2}}\sigma_X^n}\exp\{-\frac{\sum_{i=1}^{n}\limits(x_i-\mu_X)^2}{2\sigma_X^2}\}\\ L(\mu_Y,\Sigma_Y)&=\prod_{i=1}^{n}\limits p(y_i)\\&=\frac{1}{(2\pi)^{\frac{n}{2}}\sigma_Y^n}\exp\{-\frac{\sum_{i=1}^{n}\limits(y_i-\mu_Y)^2}{2\sigma_Y^2}\}\\ L(\rho_{XY})&=\frac{1}{(2\pi)^{\frac{n}{2}}(1-\rho_{XY}^2)^{\frac{n}{2}}\sigma_X^n}\cdot\\&\exp\{-\frac{1}{2(1-\rho_{XY}^2)}\sum_{i=1}^{n}\limits[\frac{(x_i-\mu_X)^2}{\sigma_X^2}-\\&\frac{2\rho_{XY}(x_i-\mu_X)(y_i-\mu_Y)}{\sigma_X\sigma_Y}+\frac{(y_i-\mu_Y)^2}{\sigma_Y^2}]\\&+\sum_{i=1}^{n}\limits\frac{(y_i-\mu_Y)^2}{2\sigma_Y^2}\} \end{array}
L(μX,ΣX)L(μY,ΣY)L(ρXY)=i=1∏np(xi)=(2π)2nσXn1exp{−2σX2i=1∑n(xi−μX)2}=i=1∏np(yi)=(2π)2nσYn1exp{−2σY2i=1∑n(yi−μY)2}=(2π)2n(1−ρXY2)2nσXn1⋅exp{−2(1−ρXY2)1i=1∑n[σX2(xi−μX)2−σXσY2ρXY(xi−μX)(yi−μY)+σY2(yi−μY)2]+i=1∑n2σY2(yi−μY)2}
对数似然函数为
l
(
μ
X
,
Σ
X
)
=
ln
L
(
μ
X
,
Σ
X
)
=
−
n
2
ln
2
π
−
n
ln
σ
X
−
∑
i
=
1
n
(
x
i
−
μ
X
)
2
2
σ
X
2
l
(
μ
Y
,
Σ
Y
)
=
ln
L
(
μ
Y
,
Σ
Y
)
=
−
n
2
ln
2
π
−
n
ln
σ
Y
−
∑
i
=
1
n
(
y
i
−
μ
Y
)
2
2
σ
Y
2
l
(
ρ
X
Y
)
=
ln
L
(
ρ
X
Y
)
=
−
n
2
ln
2
π
−
n
2
ln
(
1
−
ρ
X
Y
2
)
−
n
ln
σ
Y
−
1
2
(
1
−
ρ
X
Y
2
)
∑
i
=
1
n
[
(
x
i
−
μ
X
)
2
σ
X
2
−
2
ρ
X
Y
(
x
i
−
μ
X
)
(
y
i
−
μ
Y
)
σ
X
σ
Y
+
(
y
i
−
μ
Y
)
2
σ
Y
2
]
+
∑
i
=
1
n
(
y
i
−
μ
Y
)
2
2
σ
Y
2
\begin{array}{rl} l(\mu_X,\Sigma_X)&=\ln L(\mu_X,\Sigma_X)=-\frac{n}{2}\ln{2\pi}\\&-n\ln\sigma_X-\frac{\sum_{i=1}^{n}\limits(x_i-\mu_X)^2}{2\sigma_X^2}\\ l(\mu_Y,\Sigma_Y)&=\ln L(\mu_Y,\Sigma_Y)=-\frac{n}{2}\ln{2\pi}\\&-n\ln\sigma_Y-\frac{\sum_{i=1}^{n}\limits(y_i-\mu_Y)^2}{2\sigma_Y^2}\\ l(\rho_{XY})&=\ln L(\rho_{XY})=-\frac{n}{2}\ln{2\pi}\\&-\frac{n}{2}\ln{(1-\rho_{XY}^2)}-n\ln \sigma_Y\\&-\frac{1}{2(1-\rho_{XY}^2)}\sum_{i=1}^{n}\limits[\frac{(x_i-\mu_X)^2}{\sigma_X^2}\\&-\frac{2\rho_{XY}(x_i-\mu_X)(y_i-\mu_Y)}{\sigma_X\sigma_Y}+\frac{(y_i-\mu_Y)^2}{\sigma_Y^2}]\\&+\sum_{i=1}^{n}\limits\frac{(y_i-\mu_Y)^2}{2\sigma_Y^2} \end{array}
l(μX,ΣX)l(μY,ΣY)l(ρXY)=lnL(μX,ΣX)=−2nln2π−nlnσX−2σX2i=1∑n(xi−μX)2=lnL(μY,ΣY)=−2nln2π−nlnσY−2σY2i=1∑n(yi−μY)2=lnL(ρXY)=−2nln2π−2nln(1−ρXY2)−nlnσY−2(1−ρXY2)1i=1∑n[σX2(xi−μX)2−σXσY2ρXY(xi−μX)(yi−μY)+σY2(yi−μY)2]+i=1∑n2σY2(yi−μY)2
由
{ d l ( μ X , Σ X ) d μ X = − 2 n μ X − 2 n X ˉ 2 σ X 2 = 0 d l ( μ Y , Σ Y ) d μ Y = − 2 n μ Y − 2 n Y ˉ 2 σ Y 2 = 0 d l ( μ X , Σ X ) d σ X = − n σ X + ∑ i = 1 n ( x i − μ X ) 2 σ X 3 = 0 d l ( μ Y , Σ Y ) d σ Y = − n σ Y + ∑ i = 1 n ( y i − μ Y ) 2 σ Y 3 = 0 d l ( ρ X Y ) d ρ X Y = 0 \left\{\begin{array}{rl} \frac{dl(\mu_X,\Sigma_X)}{d\mu_X}&=-\frac{2n\mu_X-2n\bar{X}}{2\sigma_X^2}=0\\ \frac{dl(\mu_Y,\Sigma_Y)}{d\mu_Y}&=-\frac{2n\mu_Y-2n\bar{Y}}{2\sigma_Y^2}=0\\ \frac{dl(\mu_X,\Sigma_X)}{d\sigma_X}&=-\frac{n}{\sigma_X}+\frac{\sum_{i=1}^{n}\limits(x_i-\mu_X)^2}{\sigma_X^3}=0\\ \frac{dl(\mu_Y,\Sigma_Y)}{d\sigma_Y}&=-\frac{n}{\sigma_Y}+\frac{\sum_{i=1}^{n}\limits(y_i-\mu_Y)^2}{\sigma_Y^3}=0\\ \frac{dl(\rho_{XY})}{d\rho_{XY}}&=0 \end{array} \right. ⎩⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎧dμXdl(μX,ΣX)dμYdl(μY,ΣY)dσXdl(μX,ΣX)dσYdl(μY,ΣY)dρXYdl(ρXY)=−2σX22nμX−2nXˉ=0=−2σY22nμY−2nYˉ=0=−σXn+σX3i=1∑n(xi−μX)2=0=−σYn+σY3i=1∑n(yi−μY)2=0=0
得
{ μ X = x ˉ μ Y = y ˉ σ X = 1 n ∑ i = 1 n ( x i − x ˉ ) 2 σ Y = 1 n ∑ i = 1 n ( y i − y ˉ ) 2 ρ X Y = ∑ i = 1 n ( x i − x ˉ ) ( y i − y ˉ ) ∑ i = 1 n ( x i − x ˉ ) 2 ∑ i = 1 n ( y i − y ˉ ) 2 \left\{\begin{array}{rl} &\mu_X=\bar{x}\\ &\mu_Y=\bar{y}\\ &\sigma_X=\sqrt{\frac{1}{n}\sum_{i=1}^{n}\limits(x_i-\bar{x})^2}\\ &\sigma_Y=\sqrt{\frac{1}{n}\sum_{i=1}^{n}\limits(y_i-\bar{y})^2}\\ &\rho_{XY}=\frac{\sum_{i=1}^{n}\limits(x_i-\bar{x})(y_i-\bar{y})}{\sqrt{\sum_{i=1}^{n}\limits(x_i-\bar{x})^2}\sqrt{\sum_{i=1}^{n}\limits(y_i-\bar{y})^2}} \end{array} \right. ⎩⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎧μX=xˉμY=yˉσX=n1i=1∑n(xi−xˉ)2σY=n1i=1∑n(yi−yˉ)2ρXY=i=1∑n(xi−xˉ)2i=1∑n(yi−yˉ)2i=1∑n(xi−xˉ)(yi−yˉ)
故
μ
X
,
μ
Y
,
σ
X
,
σ
Y
,
ρ
X
Y
\mu_X,\mu_Y,\sigma_X,\sigma_Y,\rho_{XY}
μX,μY,σX,σY,ρXY的最大似然估计量为
μ
^
X
=
X
ˉ
,
μ
^
Y
=
Y
ˉ
,
\widehat{\mu}_X=\bar{X},\widehat{\mu}_Y=\bar{Y},
μ
X=Xˉ,μ
Y=Yˉ,
σ
^
X
=
1
n
∑
i
=
1
n
(
X
i
−
X
ˉ
)
2
,
\widehat{\sigma}_X=\sqrt{\frac{1}{n}\sum_{i=1}^{n}\limits(X_i-\bar{X})^2},
σ
X=n1i=1∑n(Xi−Xˉ)2,
σ
^
Y
=
1
n
∑
i
=
1
n
(
Y
i
−
Y
ˉ
)
2
,
\widehat{\sigma}_Y=\sqrt{\frac{1}{n}\sum_{i=1}^{n}\limits(Y_i-\bar{Y})^2},
σ
Y=n1i=1∑n(Yi−Yˉ)2,
ρ
^
X
Y
=
∑
i
=
1
n
(
X
i
−
X
ˉ
)
(
Y
i
−
Y
ˉ
)
∑
i
=
1
n
(
X
i
−
X
ˉ
)
2
∑
i
=
1
n
(
Y
i
−
Y
ˉ
)
2
.
\widehat{\rho}_{XY}=\frac{\sum_{i=1}^{n}\limits(X_i-\bar{X})(Y_i-\bar{Y})}{\sqrt{\sum_{i=1}^{n}\limits(X_i-\bar{X})^2}\sqrt{\sum_{i=1}^{n}\limits(Y_i-\bar{Y})^2}}.
ρ
XY=i=1∑n(Xi−Xˉ)2i=1∑n(Yi−Yˉ)2i=1∑n(Xi−Xˉ)(Yi−Yˉ).
2.19设 X 1 , ⋯ , X n X_1,\cdots,X_n X1,⋯,Xn为取自具有指数分布族 { p ( x ; λ ) = λ − 1 e − x / λ I ( 0 , ∞ ) ( x ) , λ > 0 } \{p(x;\lambda)=\lambda^{-1}e^{-x/\lambda}I_{(0,\infty)}(x),\lambda>0\} {p(x;λ)=λ−1e−x/λI(0,∞)(x),λ>0}的简单随机样本。验证 λ \lambda λ的无偏估计 λ ^ = X ˉ \widehat{\lambda}=\bar{X} λ =Xˉ既是 λ \lambda λ的矩法估计,又是 λ \lambda λ的最大似然估计。说明在形为 c X ˉ c\bar{X} cXˉ的 λ \lambda λ的估计量中,在均方误差准则下存在优于 λ ^ \widehat{\lambda} λ 的估计。
参考答案:
总体
X
X
X的密度函数为
p
(
x
;
λ
)
=
λ
−
1
e
−
x
/
λ
I
(
0
,
∞
)
(
x
)
,
λ
>
0.
p(x;\lambda)=\lambda^{-1}e^{-x/\lambda}I_{(0,\infty)}(x),\lambda>0.
p(x;λ)=λ−1e−x/λI(0,∞)(x),λ>0.
总体矩:
μ
1
=
E
X
=
∫
0
+
∞
x
λ
−
1
e
−
x
/
λ
d
x
=
λ
,
\mu_1=EX=\int_{0}^{+\infty}x\lambda^{-1}e^{-x/\lambda}dx=\lambda,
μ1=EX=∫0+∞xλ−1e−x/λdx=λ,
样本矩:
A
1
=
1
n
∑
i
=
1
n
X
i
=
X
ˉ
.
A_1=\frac{1}{n}\sum_{i=1}^{n}X_i=\bar{X}.
A1=n1i=1∑nXi=Xˉ.
令
μ
1
=
A
1
,
\mu_1=A_1,
μ1=A1,则
λ
=
X
ˉ
.
\lambda=\bar{X}.
λ=Xˉ.
即
λ
\lambda
λ的矩法估计量
λ
^
=
X
ˉ
.
\widehat{\lambda}=\bar{X}.
λ
=Xˉ.
设 x 1 , ⋯ , x n x_1,\cdots,x_n x1,⋯,xn是对应 X 1 , ⋯ , X n X_1,\cdots,X_n X1,⋯,Xn的样本值。
总体 X X X的密度函数为
p
(
x
;
λ
)
=
λ
−
1
e
−
x
/
λ
I
(
0
,
∞
)
(
x
)
,
λ
>
0.
p(x;\lambda)=\lambda^{-1}e^{-x/\lambda}I_{(0,\infty)}(x),\lambda>0.
p(x;λ)=λ−1e−x/λI(0,∞)(x),λ>0.
似然函数为
L
(
λ
)
=
{
1
λ
n
e
−
∑
i
=
1
n
x
i
λ
,
x
i
>
0
0
,
x
i
≤
0
(
i
=
1
,
⋯
,
n
)
L(\lambda)=\left\{\begin{array}{rl} \frac{1}{\lambda^n}e^{-\frac{\sum_{i=1}^{n}x_i}{\lambda}},x_i>0 \\ 0\qquad,x_i\le 0 \end{array}\right. (i=1,\cdots,n)
L(λ)={λn1e−λ∑i=1nxi,xi>00,xi≤0(i=1,⋯,n)
对数似然函数为
l
(
λ
)
=
ln
L
(
λ
)
=
−
n
ln
λ
−
∑
i
=
1
n
x
i
λ
,
l(\lambda)=\ln L(\lambda)=-n\ln\lambda-\frac{\sum_{i=1}^{n}x_i}{\lambda},
l(λ)=lnL(λ)=−nlnλ−λ∑i=1nxi,
x
i
>
0
(
i
=
1
,
⋯
,
n
)
x_i>0(i=1,\cdots,n)
xi>0(i=1,⋯,n)
令
d
l
(
λ
)
d
λ
=
−
n
λ
+
∑
i
=
1
n
x
i
λ
2
=
0
,
\frac{dl(\lambda)}{d\lambda}=-\frac{n}{\lambda}+\frac{\sum_{i=1}^{n}x_i}{\lambda^2}=0,
dλdl(λ)=−λn+λ2∑i=1nxi=0,
得
λ
=
∑
i
=
1
n
x
i
n
=
x
ˉ
.
\lambda=\frac{\sum_{i=1}^{n}x_i}{n}=\bar{x}.
λ=n∑i=1nxi=xˉ.
则
λ
\lambda
λ最大似然估计量为
λ
^
M
L
E
=
X
ˉ
.
\widehat{\lambda}_{MLE}=\bar{X}.
λ
MLE=Xˉ.
设
λ
^
1
=
c
X
ˉ
,
\widehat{\lambda}_1=c\bar{X},
λ
1=cXˉ,有
E
(
λ
^
1
)
=
E
(
c
X
ˉ
)
=
c
E
(
X
ˉ
)
=
c
λ
,
V
a
r
(
λ
^
1
)
=
V
a
r
(
c
X
ˉ
)
=
c
2
V
a
r
(
X
ˉ
)
=
c
2
λ
2
n
\begin{array}{rl} & E(\widehat{\lambda}_1)=E(c\bar{X})=cE(\bar{X})=c\lambda,\\ & Var(\widehat{\lambda}_1)=Var(c\bar{X})=c^2Var(\bar{X})=\frac{c^2\lambda^2}{n} \end{array}
E(λ
1)=E(cXˉ)=cE(Xˉ)=cλ,Var(λ
1)=Var(cXˉ)=c2Var(Xˉ)=nc2λ2
则
M
S
E
(
X
ˉ
)
=
V
a
r
(
X
ˉ
)
+
(
E
(
X
ˉ
)
−
λ
)
2
=
λ
2
n
,
M
S
E
(
λ
^
1
)
=
V
a
r
(
λ
^
1
)
+
(
E
(
λ
^
1
)
−
λ
)
2
=
c
2
λ
2
n
+
(
c
λ
−
λ
)
2
=
[
n
+
1
n
(
c
−
n
n
+
1
)
2
+
1
n
+
1
]
λ
2
.
\begin{array}{rl} MSE(\bar{X})&=Var(\bar{X})+(E(\bar{X})-\lambda)^2\\&=\frac{\lambda^2}{n},\\ MSE(\widehat{\lambda}_1)&=Var(\widehat{\lambda}_1)+(E(\widehat{\lambda}_1)-\lambda)^2\\&=\frac{c^2\lambda^2}{n}+(c\lambda-\lambda)^2\\&=[\frac{n+1}{n}(c-\frac{n}{n+1})^2+\frac{1}{n+1}]\lambda^2. \end{array}
MSE(Xˉ)MSE(λ
1)=Var(Xˉ)+(E(Xˉ)−λ)2=nλ2,=Var(λ
1)+(E(λ
1)−λ)2=nc2λ2+(cλ−λ)2=[nn+1(c−n+1n)2+n+11]λ2.
当
c
=
n
n
+
1
c=\frac{n}{n+1}
c=n+1n时,
λ
^
1
=
n
n
+
1
X
ˉ
\widehat{\lambda}_1=\frac{n}{n+1}\bar{X}
λ
1=n+1nXˉ的均方误差
M
S
E
(
λ
^
1
)
=
1
n
+
1
λ
2
MSE(\widehat{\lambda}_1)=\frac{1}{n+1}\lambda^2
MSE(λ
1)=n+11λ2小于
X
ˉ
\bar{X}
Xˉ的均方误差
M
S
E
(
X
ˉ
)
=
λ
2
n
,
MSE(\bar{X})=\frac{\lambda^2}{n},
MSE(Xˉ)=nλ2,
故在形为 c X ˉ c\bar{X} cXˉ的 λ \lambda λ的估计量中,在均方误差准则下存在优于 λ ^ \widehat{\lambda} λ 的估计。
如果您在看完以后,无论是题目还是参考解答过程有什么问题,希望不吝指出!如果有更好的解题思路或者好的问题分享,欢迎通过后台发送给我们!