STAT 5204 Quiz 1
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
STAT 5204 Quiz 1
Problem 1
(a) Show that the Beta(α, β) distribution is a conjugate prior for iid data X1, . . . , Xn from a Bernoulli(p) distribution with unknown parameter p. Explicitly state the parameters of the posterior distribution. (b) With the above setup, find the Bayes estimator pˆB of p under squared error loss. Next, express pˆB as a weighted average of the sample mean and the prior mean, i.e. pˆB = Wn n + (1 − Wn)µ where µ is the mean of the Beta(α, β) prior. What happens to pˆB as n → ∞?
Problem 2
The number of subway trains stopping at 116th Street every hour follows a Poisson distribution with unknown parameter λ. I hung out on the platform for 7 hours and counted a total of 45 trains. Now I want to estimate λ using the Bayesian method. I put a Gamma(5, 7) prior on λ (using the shape and scale parametrization). What is my Bayes estimate of λ under squared error loss?
Hint: The Gamma distribution is a conjugate prior for a Poisson likelihood.
Problem 3
Denote the parameter by θ = (θ1, . . . , θs) ∈ Rs, which may be multi-dimensional if s > 1. A family of distributions parametrized by θ is said to belong to an exponential family if the probability density function (or probability mass function, for discrete distributions) can be written as:
f (x | θ) = h (x) exp ηi (θ) Ti (x) − A (θ)!
(a) Show that the normal distribution with unknown mean µ ∈ R and unknown variance σ2 > 0 belongs to an exponential family.
(b) Show that the uniform distribution on 0, θ , which has density f (x | θ) = 1 ✶0≤x≤θ, does not belong to an exponential family. θ
Hint: What can you say about the support of an exponential family?
2022-03-07