Chapter 2.7

Sufficient Statistics and Optimal Estimators

[8/100%]
Show
LaTeX
Problem 2.7.1

Give an example of a parametric family of distributions P\mathcal{P} and a nontrivial, incomplete sufficient statistic S(X1,…,Xn)S(X_1, \ldots , X_n), where X1,…,XnX_1, \ldots , X_n is a sample from an unknown distribution P∈P\mathbb {P} \in \mathcal{P}, such that the dimension of the statistic SS equals 1

?
Problem 2.7.2

Prove that in the model

N(θ,γθ2),(θ,γ)∈R×(0,∞), N(\theta , \gamma \theta ^2), \quad (\theta , \gamma ) \in \mathbb {R} \times (0, \infty ),

the statistic (X‾,s2)(\overline{X}, s^2) is complete but not sufficient.

?
Problem 2.7.3

Find sufficient statistics for the following parametric distributions:

?
(a)

N(a,σ2)N(a, \sigma^2);

(b)

Γ(α,λ)\Gamma (\alpha , \lambda );

(c)

R(a,b)R(a,b);

(d)

Pois(λ)\mathrm{Pois}(\lambda );

(e)

Ber(p)\mathrm{Ber}(p);

(f)

Geom(p)\mathrm{Geom}(p).

Problem 2.7.4

Find the optimal estimator of θ>0\theta > 0 from a sample of the distribution

?
(a)

N(θ,1)N(\theta , 1);

(b)

R(0,θ)R(0, \theta );

(c)

Pois(θ)\mathrm{Pois}(\theta );

(d)

Bern(θ),  θ∈(0,1)\mathrm{Bern}(\theta ), \; \theta \in (0,1).

Problem 2.7.5

Let X1,…,XnX_1, \ldots , X_n be a sample from a normal distribution with parameters (a,σ2),  a∈R,  σ>0(a,\sigma^2), \; a \in \mathbb {R}, \; \sigma > 0. Find the optimal estimator of the parameter θ=(a,σ2)\theta = (a, \sigma^2).

?
Problem 2.7.6

Let X1,X2,…,XnX_1, X_2, \dots , X_n be a sample from Exp(θ)\mathrm{Exp}(\theta ). Find the optimal estimators for θ\theta and for θ\sqrt{\theta }.

?
Problem 2.7.7

Let X1,X2,…,XnX_1, X_2, \dots , X_n be a sample from N(0,θ2)N(0, \theta^2). Find the optimal estimator for θ\theta.

?
Problem 2.7.8

Let X1,…,XnX_1, \dots , X_n be a sample from a Poisson distribution with parameter θ>0\theta > 0. Find

E[X12  ∣  ∑i=1nXi]. \mathbb {E}\left[X_1^2 \; \mid \; \sum _{i=1}^n X_i\right].
?