Scribd Logo
Upload

Welcome to Scribd!

  • 37 views

    Tutorial 4

    Uploaded by

    Amrit Raj
    The document provides solutions to tutorial problems regarding parameter estimation methods. For problem 8.10.12, the solution recommends choosing maximum likelihood estimation (MLE) over method of moments estimation. MLE provides more accurate estimates as simulations show the sampling distributions of MLEs are less dispersed. For problem 8.10.21, the document finds the method of moments and MLE of parameter θ and shows the minimum of the observations is a sufficient statistic for θ. It also solves for the likelihood and MLE of θ given coin toss outcomes in problem 8.10.31.

    Copyright:

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

    Tutorial 4

    Uploaded by

    Amrit Raj
    37 views7 pages
    The document provides solutions to tutorial problems regarding parameter estimation methods. For problem 8.10.12, the solution recommends choosing maximum likelihood estimation (MLE) over method of moments estimation. MLE provides more accurate estimates as simulations show the sampling distributions of MLEs are less dispersed. For problem 8.10.21, the document finds the method of moments and MLE of parameter θ and shows the minimum of the observations is a sufficient statistic for θ. It also solves for the likelihood and MLE of θ given coin toss outcomes in problem 8.10.31.

    Original Description:

    Dse

    Original Title

    tutorial4

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document provides solutions to tutorial problems regarding parameter estimation methods. For problem 8.10.12, the solution recommends choosing maximum likelihood estimation (MLE) over method of moments estimation. MLE provides more accurate estimates as simulations show the sampling distributions of MLEs are less dispersed. For problem 8.10.21, the document finds the method of moments and MLE of parameter θ and shows the minimum of the observations is a sufficient statistic for θ. It also solves for the likelihood and MLE of θ given coin toss outcomes in problem 8.10.31.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    37 views7 pages

    Tutorial 4

    Uploaded by

    Amrit Raj
    The document provides solutions to tutorial problems regarding parameter estimation methods. For problem 8.10.12, the solution recommends choosing maximum likelihood estimation (MLE) over method of moments estimation. MLE provides more accurate estimates as simulations show the sampling distributions of MLEs are less dispersed. For problem 8.10.21, the document finds the method of moments and MLE of parameter θ and shows the minimum of the observations is a sufficient statistic for θ. It also solves for the likelihood and MLE of θ given coin toss outcomes in problem 8.10.31.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 7

    TUTORIAL 4 SOLUTIONS

    #8.10.12 Suppose that you had to choose


    either the method of moments estimates or
    the maximum likelihood estimates in Exam-
    ple C of Section 8.4 and C of Section 8.5.
    Which would you choose and why?
    Solution We shall focus on Example C
    of Section 8.4 as the same reasons apply to
    Example C of Section 8.5.
    The method of moments estimates for the
    parameters λ and α are available in closed-
    form:
    X̄
    λ̂ = 2 ,
    σ̂
    X̄ 2
    α̂ = 2 .
    σ̂
    This is typical for method of moments esti-
    mates and is an attractive property of method
    of moments estimates.
    1
    On the other hand, the mle’s for λ and α
    are not available in closed-form.
    However it can be shown by simulations
    (and also theoretically) that the sampling
    distributions of the two mle’s are substan-
    tially less dispersed than those of the method
    of moments estimates.
    This means that the two mle’s are gener-
    ally more accurate than the method of mo-
    ments estimates for λ and α.

    2
    #8.10.21 Suppose that X1, . . . , Xn are
    i.i.d. with density function
    f (x|θ) = e−(x−θ) , x ≥ θ,
    and f (x|θ) = 0 otherwise.
    a. Find the method of moments estimate of
    θ.
    b. Find the mle of θ. (Hint: Be careful,
    don’t differentiate before thinking. For
    what values of θ is the likelihood posi-
    tive?)
    c. Find a sufficient statistic for θ.

    Solution
    a. Let µ = E(X1). Then
    Z ∞
    µ= xe−(x−θ) dx
    Zθ ∞
    = (x + θ)e−xdx
    0
    = θ + 1.
    3
    Thus θ = µ − 1 and the method of moments
    estimate of θ is
    θ̂ = µ̂1 − 1 = X̄ − 1.

    b. Note that the density function f (x|θ)


    is not differentiable at x = θ.
    Also f (x|θ) is strictly decreasing for x ≥ θ
    and equals 0 for x < θ.
    The likelihood function is
    n
    Y
    lik(θ) = e−(xi−θ)
    i=1
    if min(x1, . . . , xn) ≥ θ, and lik(θ) = 0 oth-
    erwise.
    Hence the maximum of lik(θ) occurs at
    θ = min(x1, . . . , xn) and we conclude that
    the mle of θ is
    θ̃ = min(X1, . . . , Xn).

    4
    c. Let I{xi ≥ θ} denote the indicator
    function of the event {xi ≥ θ}.
    I.e., I{xi ≥ θ} = 1 if xi ≥ θ and I{xi ≥
    θ} = 0 if xi < θ.
    Then the joint pdf of X is given by
    f (x|θ)
    Yn
    = e−(xi−θ) I{xi ≥ θ}
    i=1P
    n
    = e − i=1 xi enθ I{min(x , . . . , xn ) ≥ θ}
    1
    = g(t, θ)h(x),
    where
    t = min(x1, . . . , xn),
    g(t, θ) = enθP
    I{min(x1, . . . , xn) ≥ θ},
    n
    h(x) = e − i=1 xi .

    It follows from the factorization theorem that


    T = min(X1, . . . , Xn) is sufficient for θ.

    5
    #8.10.31 George spins a coin three times
    and observes no heads. He then gives the
    coin to Hilary. She spins it until the first
    head occurs, and ends up spinning it four
    times total. Let θ denote the probability
    that the coin comes up heads.
    a. What is the likelihood of θ?
    b. What is the MLE of θ?

    Solution
    a. The likelihood function is
    lik(θ) = P (heads)[P (tails)]6
    = θ(1 − θ)6.

    6
    b. The loglikelihood is
    l(θ) = log(θ) + 6 log(1 − θ).
    Differentiating l(θ) wrt θ and then solving
    for l0(θ) = 0, we get
    0 1 6
    (1) l (θ) = − = 0.
    θ 1−θ
    The solution for (1) is θ = 1/7. Hence we
    conclude that the MLE of θ is
    θ̂ = 1/7.

    You might also like