1 求解步骤
1.写出似然函数：
- 离散型：
L ( θ ) = L ( x 1 , x 2 , ⋯ , x n ; θ 1 , θ 2 , ⋯ , θ m ) = ∏ i = 1 n P ( x i ; θ 1 , θ 2 , ⋯ , θ m ) L (\theta) = L (x_1, x_2, \cdots, x _n; \theta_1, \theta_2, \cdots, \theta_m) = \prod_{i = 1}^n P (x_i; \theta_1, \theta_2, \cdots, \theta_m) L(θ)=L(x1,x2,⋯,xn;θ1,θ2,⋯,θm)=i=1∏nP(xi;θ1,θ2,⋯,θm) - 连续型：
L ( θ ) = L ( x 1 , x 2 , ⋯ , x n ; θ 1 , θ 2 , ⋯ , θ m ) = ∏ i = 1 n f ( x i ; θ 1 , θ 2 , ⋯ , θ m ) L (\theta) = L (x_1, x_2, \cdots, x _n; \theta_1, \theta_2, \cdots, \theta_m) = \prod_{i = 1}^n f (x_i; \theta_1, \theta_2, \cdots, \theta_m) L(θ)=L(x1,x2,⋯,xn;θ1,θ2,⋯,θm)=i=1∏nf(xi;θ1,θ2,⋯,θm)
- 取对数 ln L \ln L lnL；
- 分别对 θ 1 , θ 2 , ⋯ , θ m \theta_1, \theta_2, \cdots, \theta_m θ1,θ2,⋯,θm求偏导 ∂ ln L ∂ θ i \frac{\partial \ln L}{\partial \theta_i} ∂θi∂lnL， i = 1 , 2 , ⋯ , m i = 1, 2, \cdots, m i=1,2,⋯,m；
- 判断方程组 ∂ ln L ∂ θ i \frac{\partial \ln L}{\partial \theta_i} ∂θi∂lnL是否有解。若有解，则其解为所求最大似然估计；若无解，则最大似然估计常在 θ i \theta_i θi的边界点上达到。
2 例题
- 设总体 X X X的分布律为 P { X = k } = p ( 1 − p ) k − 1 , k = 1 , 2 , ⋯ P \{ X = k \} = p (1 - p)^{k - 1}, k = 1, 2, \cdots P{X=k}=p(1−p)k−1,k=1,2,⋯，其中 p p p为未知参数，且 X 1 , X 2 , ⋯ , X n X_1, X_2, \cdots, X_n X1,X2,⋯,Xn为来自总体 X X X的简单随机样本，求参数 p p p的矩估计量和极大似然估计量。
- 矩估计：
E ( x ) = ∑ k = 1 ∞ k ⋅ p ( 1 − p ) k − 1 = p ( ∑ k = 1 ∞ x k ) ′ ∣ x = 1 − p = p ( x 1 − x ) ′ ∣ x = 1 − p = p 1 ( 1 − x ) 2 ∣ x = 1 − p = 1 p = x ‾ \begin{aligned} E (x) &= \sum_{k = 1}^{\infty} k \cdot p (1 - p)^{k - 1}\\ &= p (\sum_{k = 1}^\infty x^k)' \Big|_{x = 1 - p}\\ &= p \left( \frac{x}{1 - x} \right)'\Big|_{x = 1 - p}\\ &= p \frac{1}{(1 - x)^2}\Big|_{x = 1 - p}\\ &=\frac{1}{p} = \overline{x} \end{aligned} E(x)=k=1∑∞k⋅p(1−p)k−1=p(k=1∑∞xk)′∣∣∣x=1−p=p(1−xx)′∣∣∣x=1−p=p(1−x)21∣∣∣x=1−p=p1=x则有 p ^ = 1 x ‾ \hat{p} = \frac{1}{\overline{x}} p^=x1 - 极大似然估计：
L ( p ) = P { X = x 1 } P { X = x 2 } ⋯ P { X = x n } = p ( 1 − p ) x 1 − 1 p ( 1 − p ) x 2 − 1 ⋯ p ( 1 − p ) x n − 1 = p n ( 1 − p ) x 1 + x 2 + ⋯ + x n − n \begin{aligned} L (p) &= P \{ X = x_1 \} P \{ X = x_2 \} \cdots P \{ X = x_n \}\\ &= p (1 - p)^{x_1 - 1} p (1 - p)^{x_2 - 1} \cdots p (1 - p)^{x_n - 1}\\ &= p^n (1 - p)^{x_1 + x_2 + \cdots + x_n - n} \end{aligned} L(p)=P{X=x1}P{X=x2}⋯P{X=xn}=p(1−p)x1−1p(1−p)x2−1⋯p(1−p)xn−1=pn(1−p)x1+x2+⋯+xn−n ln L ( p ) = n ln p + ( x 1 + x 2 + ⋯ + x n − n ) ln ( 1 − p ) \ln L (p) = n \ln p + (x_1 + x_2 + \cdots + x_n - n) \ln (1 - p) lnL(p)=nlnp+(x1+x2+⋯+xn−n)ln(1−p) ∂ ln L ( p ) ∂ p = n p − x 1 + x 2 + ⋯ + x n − n 1 − p = 0 \frac{\partial \ln L (p)}{\partial p} = \frac{n}{p} - \frac{x_1 + x_2 + \cdots + x_n - n}{1 - p} = 0 ∂p∂lnL(p)=pn−1−px1+x2+⋯+xn−n=0 n − n p = p ( x 1 + x 2 + ⋯ + x n − n ) n - np = p (x_1 + x_2 + \cdots + x_n - n) n−np=p(x1+x2+⋯+xn−n) p = 1 x 1 + x 2 + ⋯ + x n n = 1 x ‾ p = \frac{1}{\frac{x_1 + x_2 + \cdots + x_n}{n}} = \frac{1}{\overline{x}} p=nx1+x2+⋯+xn1=x1
- 设总体
X
X
X的概率密度为
f ( x ; θ ) = { θ 2 x 3 e − θ x , x > 0 , 0 , 其他 , f (x; \theta) = \left\{ \begin{aligned} &\frac{\theta^2}{x^3} e^{- \frac{\theta}{x}}, &x > 0,\\ &0, & \text{其他}, \end{aligned} \right. f(x;θ)=⎩⎨⎧x3θ2e−xθ,0,x>0,其他,其中 θ \theta θ为未知参数且大于零， X 1 , X 2 , ⋯ , X n X_1, X_2, \cdots, X_n X1,X2,⋯,Xn为来自总体 X X X的简单随机样本。求参数 θ \theta θ的矩估计量和极大似然估计量。
- 矩估计量：
E ( x ) = ∫ 0 + ∞ x ⋅ θ 2 x 3 e − θ x d x = θ E (x) = \int_0^{+ \infty} x \cdot \frac{\theta^2}{x^3} e^{- \frac{\theta}{x}} d x = \theta E(x)=∫0+∞x⋅x3θ2e−xθdx=θ - 最大似然估计量：
L ( θ ) = ∏ i = 1 n f ( x i ; θ ) L (\theta) = \prod_{i = 1}^n f (x_i; \theta) L(θ)=i=1∏nf(xi;θ)
θ ^ = 2 n ∑ i = 1 n 1 X i \hat{\theta} = \frac{2 n}{\sum_{i = 1}^n \frac{1}{X_i}} θ^=∑i=1nXi12n
参考文献：
【1】最大似然估计【小元老师】考研数学，概率论与数理统计
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
