Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
24 views

Collection of Assignments

This document contains exercises related to statistics and probability. It includes exercises on finding unbiased estimators, the method of moments estimator, consistency of the empirical distribution function, and properties of kernel density estimators. It also covers finding sufficient statistics for various distributions and showing a statistic is not sufficient.

Uploaded by

Kumar Nilesh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
24 views

Collection of Assignments

This document contains exercises related to statistics and probability. It includes exercises on finding unbiased estimators, the method of moments estimator, consistency of the empirical distribution function, and properties of kernel density estimators. It also covers finding sufficient statistics for various distributions and showing a statistic is not sufficient.

Uploaded by

Kumar Nilesh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Statistics – I: Course – MA3205

Exercise Set 3

Spring 2019

iid
Exercise 1. Let X1 , X2 , . . . , Xn ∼ Fθ for θ ∈ Θ. The corresponding pdf/pmf is denoted by fθ . For each
of the following models, find an unbiased estimator of the specified parameter of interest.
(a) fθ is the N(µ, σ 2 ) density, and the parameter of interest is σ.
[Hint: Think in terms of s, and recall that (n − 1)s2 ∼ σ 2 χ2(n−1) .]
(b) fθ (x) = 3x√2 θ −3 exp{−(x/θ)3 }, x ≥ 0, and the parameter of interest if θ.

(c) fθ (x) = (x  2π)−1exp{−(ln x − µ)2 /2}, x ≥ 0, and the parameter of interest is µ.


K N −K N

(d) fθ (x) = x n−x / n , max(0, n + K − N ) ≤ x ≤ min(K, n) with N and n fixed, and the parameter
of interest is K.
[Hint: When calulating the expectation in (d), use combinatorial properties and try to reduce the
expression to a form similar to the pmf.]

Exercise 2. Let X1 , X2 , . . . , Xn be an iid sample from the distribution with pdf

(ln x − µ)2
 
1
fθ (x) = √ exp − , x ≥ 0,
xσ 2π 2σ 2

where θ = (µ, σ 2 )T . Find the method of moments estimator of θ.

Exercise 3. Let X1 , X2 , . . . , Xn be an iid sample from a distribution with an unknown cdf F (even the
functional form of F is unknown). For x ∈ R, define
n
1X
Fn (x) = 1(Xi ≤ x).
n
i=1

P
Show that Fn (x) is a consistent estimator of F (x), i.e., |Fn (x) − F (x)| → 0 as n → ∞.
Suppose that F is known to be a continuous cdf with density f . An estimator of f (x) based on
X1 , X2 , . . . , Xn is given by
Fn (x + hn ) − Fn (x − hn )
fn (x) = .
2hn
(a) Show that 2nhn fn (x) has a Binomial distribution with parameters n and F (x + hn ) − F (x − hn ).
(b) Show that E[fn (x)] → f (x) as n → ∞ provided that hn → 0 as n → ∞.
(c) Show that Var[fn (x)] → 0 as n → ∞ provided that hn → 0 and nhn → ∞ as n → ∞.
P
(d) Use (b) and (c) to show that |fn (x) − f (x)| → 0 as n → ∞.
[Hard] Suppose it is additionally known that F is a continuous cdf on [0, 1]. In that case, show that
P
sup |Fn (x) − F (x)| → 0 as n → ∞?
x∈[0,1]

[Hint: Use the fact that [0, 1] is a compact set and that the functions are cdfs (hence also monotone) to
go from pointwise convergence to uniform convergence].

Exercise 4. Find a sufficient statistic (may be vector-valued) in each of the following cases based on a
random sample of size n.

1
(a) Beta(a, b), where (i) a is known but b is unknown, (ii) b is known but a is unknown, and (iii) both a
and b are unknown. Here a, b > 0.
(b) Gamma(a, b), where (i) a is known but b is unknown, (ii) b is known but a is unknown, and (iii) both
a and b are unknown. Here a, b > 0.
(c) Unif(a, b), where (i) a is known but b is unknown, (ii) b is known but a is unknown, and (iii) both a
and b are unknown. Here a < b ∈ R.
(d) the pdf is the same as in Exercise (2), where (i) µ is known but σ is unknown, (ii) σ is known but µ
is unknown, and (iii) both µ and σ are unknown. Here µ ∈ R and σ > 0.
(e) the pmf is fθ (x) = 2(1 − 2−1/θ )2−x/θ , x = θ, θ + 1, . . . and θ > 0.
(f) N(ασ, σ 2 ), where α is a known real number and σ > 0 is the parameter.
(g) Unif(θ − 0.5, θ + 0.5), where θ ∈ R.

Exercise 5. Let X1 , X2 , . . . , Xn be an iid sample from a discrete distribution Fθ , where θ ∈ Θ ⊆ Rk .


Let T (X1 , X2 , . . . , Xn ) = (X1 , X2 , . . . , Xm ), where m < n. Show that T is not sufficient for θ.
[Hint: Use the definition of sufficiency.]

You might also like