Unformatted text preview: nsity of θ, i.e., the unnormalized
function. The plot in (e), on the other hand, corresponds to the normalized posterior density
function. These plots suggest that the most likely value of θ is 0.57. 3 0.7
0.6 0.06 0.2 0.3 p(θ∑ y=57) 0.4 0.5 0.05
0.04
0.03 p(∑ y=57θ) 0.02 0.1 0.01 0.0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 1.0 0.8 1.0 0 0.00 2 4 0.04 p(θ∑ y=57) 6 0.06 8 0.08 θ 0.02 p(θ)*p(∑ y=57θ) 0.8 0.6 θ 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 θ θ Figure 1: The plots correspond to: p( i Yi = y θ) versus θ for (b); p(θ
p(θ) × p( n Yi = 57θ) versus θ for (d); p(θ n Yi = 57) for (e)
i=1
i=1 i Yi = 57) versus θ for (c); 3. Suppose the number of defects in a 1200foot roll of magnetic recording tape has a Poisson distribution
with mean θ. Let the prior distribution of θ be Gamma(3, 1). When ﬁve rolls of this tape are selected
at random and inspected, the number of defects found on the rolls are 2, 2, 6, 0, and 3.
(a) Determine the posterior distribution of θ.
The likelihood function is given by
L( θ ) = n
e − θ θ yi yi ! i=1 = e −n θ θ
n i=1 yi i=1 yi ! and the prior distribution of θ is
π (θ ) = n ba a − 1 − b θ
θe
Γ(a) Thus, the posterior distribution of θ is
p( θ  y ) ∝ e − n θ θ
4 n i=1 yi a − 1 − b θ θ e =θ n i=1 yi + a − 1 − θ (n + b ) e We recognize the kernel of a Gamma density
θy ∼ Gamma(a + n
yi , b + n) i=1 Note that the posterior mean of θ can be presented as the weighted average of the MLE and
the pior mean:
n
a + n yi
n
ba
i=1
i=1 yi
E [θ y ] =
=
+
b+n
b+n
n
b+nb
For the observed data, n = 5 and
thus, 5 i=1 yi = 13, and the hyperparameters were set to a = 3, b = 1, θy ∼ Gamma(16, 6)
Some prior and posterior summary statistics of θ are:
E [θ ] =
E [θ y ] = a
=3
b Mode(θ) =...
View
Full
Document
This note was uploaded on 02/22/2013 for the course MATH 604 taught by Professor Tadasee during the Spring '13 term at Georgetown.
 Spring '13
 Tadasee

Click to edit the document details