Scribd Logo
Upload

Welcome to Scribd!

  • 42 views

    Fall 2011

    Uploaded by

    Robinson Ortega Meza
    This document contains solutions to probability and statistics problems: 1. It finds the mean and variance of the sample mean Ybar of a random variable Y with a particular density function. It shows that theta-hat = (3/2)Ybar is an unbiased estimator of the parameter theta, and states the Cramer-Rao inequality for this situation. It shows that the variance of theta-hat violates the Cramer-Rao inequality. 2. It shows that if X has a normal distribution and g is a differentiable function, then the expected value of g(X)(X - theta) equals the variance times the expected value of the derivative of g. It also shows that if X has

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Fall 2011

    Uploaded by

    Robinson Ortega Meza
    42 views2 pages
    This document contains solutions to probability and statistics problems: 1. It finds the mean and variance of the sample mean Ybar of a random variable Y with a particular density function. It shows that theta-hat = (3/2)Ybar is an unbiased estimator of the parameter theta, and states the Cramer-Rao inequality for this situation. It shows that the variance of theta-hat violates the Cramer-Rao inequality. 2. It shows that if X has a normal distribution and g is a differentiable function, then the expected value of g(X)(X - theta) equals the variance times the expected value of the derivative of g. It also shows that if X has

    Original Title

    2011F

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document contains solutions to probability and statistics problems: 1. It finds the mean and variance of the sample mean Ybar of a random variable Y with a particular density function. It shows that theta-hat = (3/2)Ybar is an unbiased estimator of the parameter theta, and states the Cramer-Rao inequality for this situation. It shows that the variance of theta-hat violates the Cramer-Rao inequality. 2. It shows that if X has a normal distribution and g is a differentiable function, then the expected value of g(X)(X - theta) equals the variance times the expected value of the derivative of g. It also shows that if X has

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    42 views2 pages

    Fall 2011

    Uploaded by

    Robinson Ortega Meza
    This document contains solutions to probability and statistics problems: 1. It finds the mean and variance of the sample mean Ybar of a random variable Y with a particular density function. It shows that theta-hat = (3/2)Ybar is an unbiased estimator of the parameter theta, and states the Cramer-Rao inequality for this situation. It shows that the variance of theta-hat violates the Cramer-Rao inequality. 2. It shows that if X has a normal distribution and g is a differentiable function, then the expected value of g(X)(X - theta) equals the variance times the expected value of the derivative of g. It also shows that if X has

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 2

    Fall 2011

    1. Let Y1 , . . . , Yn be a random sample from the density p(y|θ) = 2y/θ2 , 0 < y < θ, where θ > 0 is an
    unknown parameter.
    Pn
    a) Let Ȳ = (1/n) i=1 Yi . Find the mean and variance of Ȳ .
    Solution. Note that
    n Z θ θ
    1X 2y 2 y 3 2
    EȲ = EYi = y 2 dy = = θ,
    n i=1 0 θ 3 θ 2 0 3
    and θ
    θ
    1 y 4
    Z
    2y 1
    EY12 = y2 2 dy = = θ2 .
    0 θ 2 θ 2 0 2
    Hence,
    n  
    1 X 1 2 2
     1 1 2 4 2 1 2
    Var(Ȳ ) = 2 Var(Yi ) = EY1 − (EY1 ) = θ − θ = θ .
    n i=1 n n 2 9 18n

    b) Show that θ̂ = (3/2)Ȳ is an unbiased estimator of θ.


    Solution. As we saw in part (a), EȲ = 23 θ. Hence,
     
    3 32
    E Ȳ = θ = θ,
    2 23

    and so, 32 Ȳ is an unbiased estimator of θ.


    c) State the Cramer-Rao inequality for the above situation.
    Solution. The likelihood function is
    n
    2n yi
    Q
    Y 2yi
    L(θ; y) = = ,
    i=1
    θ2 θ2n

    and so, the log-likelihood function is


    n
    X
    log L(θ; y) = n log 2 + log yi − 2n log θ.
    i=1

    Hence,
    d 2n d2 2n
    log L(θ; y) = − , and 2 log L(θ; y) = 2 ,
    dθ θ dθ θ
    and so, the Fisher information is
    2n
    I(θ) = − 2 .
    θ
    Thus, if the regularity conditions hold, Cramer-Rao inequality states that for any unbiased esti-
    mator W (Y ),
    θ2
    Var(W (Y )) ≥ − .
    2n
    d) Show that Var(θ̂) violates the Cramer-Rao inequality. Explain why.
    2. a) Let X have normal distribution with mean θ and variance σ 2 , and let g be a differentiable function
    satisfying E|g 0 (X)| < ∞. Show that

    E [g(X)(X − θ)] = σ 2 Eg 0 (X).

    [HINT: use integration by parts of Fubini’s theorem.]


    Solution. We have
    Z ∞
    (x − θ)2
     
    1
    E [g(X)(X − θ)] = √ g(x)(x − θ) exp − dx
    2πσ 2 −∞ 2σ 2
    :0Z ∞
     
      ∞
    2
      2
    
    1 (x −
    θ) (x − θ)
    g 0 (x) exp −
    
    σ 2  g(x) exp −

    = −√ − dx
     
     2 2
    2πσ 2   2σ
    −∞ −∞ 2σ
    
    Z ∞
    (x − θ)2
     
    1
    = σ2 √ g 0 (x) exp − dx = σ 2 Eg 0 (X),
    2πσ 2 −∞ 2σ 2

    where the marked quantity vanishes to 0 since E|g 0 (X)| < ∞.


    b) Let g(x) be a function with −∞ < Eg(X) < ∞ and g(−1) is finite. If X has a Poisson distribution
    with mean λ, show that
    E (λg(X)) = E (X(g(X − 1))) .
    Solution. We have
    ∞ ∞
    X e−λ λx X e−λ λx+1
    E (λg(X)) = λg(x) = g(x) (x + 1)
    x=0
    x! x=0
    (x + 1)!
    ∞
    (∗) X e−λ λx
    = xg(x − 1)
    x=0
    x!
    = E (X (g(X − 1))) ,

    where the index shift in (∗) works because g(−1) is finite.

    You might also like