|
|
|
@ -395,7 +395,27 @@ $$
|
|
|
|
|
|
|
|
|
|
> 第一步构造函数;第二步取对数,对数后的值容易取且极值点还是那个位置;第三步求偏导;得到θ
|
|
|
|
|
|
|
|
|
|
![1603626465653](assets/1603626465653.png)
|
|
|
|
|
**求一个具体的值**:
|
|
|
|
|
|
|
|
|
|
![1603626480659](assets/1603626480659.png)
|
|
|
|
|
设 X 服从参数 λ(λ>0) 的泊松分布,x1,x2,...,xn 是来自 X 的一个样本值,求λ的极大似然估计值
|
|
|
|
|
|
|
|
|
|
- $$
|
|
|
|
|
因为X的分布律为P\{X=x\}=\frac{\lambda^x}{x!}e^{-\lambda},(x=0,1,2,...,n)
|
|
|
|
|
$$
|
|
|
|
|
|
|
|
|
|
- $$
|
|
|
|
|
所以\lambda的似然函数为L(\lambda)=\prod^n_{i=1}(\frac{\lambda^{x_i}}{x_i!}e^{-\lambda})=e^{-n\lambda}\frac{\lambda^{\sum^n_{i=1}x_i}}{\prod^n_{i=1}(x_i!)},
|
|
|
|
|
$$
|
|
|
|
|
|
|
|
|
|
- $$
|
|
|
|
|
lnL(\lambda)=-n\lambda+(\sum^n_{i=1}x_i)ln\lambda-\sum^n_{i=1}(x_i!),
|
|
|
|
|
$$
|
|
|
|
|
|
|
|
|
|
- $$
|
|
|
|
|
令\frac{d}{d\lambda}lnL(\lambda)=-n+\frac{\sum^n_{i=1}x_i}{\lambda}=0
|
|
|
|
|
$$
|
|
|
|
|
|
|
|
|
|
- $$
|
|
|
|
|
解得\lambda的极大似然估计值为\hat{\lambda}=\frac{1}{n}\sum_{i=1}^nx_i=\overline{x}
|
|
|
|
|
$$
|
|
|
|
|
|
|
|
|
|