Scribd
Upload
190 views3 pages

Homework8 Ans

1. The maximum likelihood estimator (MLE) for θ is θ̂ = (Σki)/n, where ki is the observed value for each sample and n is the sample size. 2. For the statistic c(Y1 + 2Y2) to be an unbiased estimator of θ1, c must equal 0.5. 3. W2 can only be an unbiased estimator of θ2 if the variance of W is 0, meaning W would have to be a constant value.

Uploaded by

Ashok Thiruvengadam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
190 views3 pages

Homework8 Ans

1. The maximum likelihood estimator (MLE) for θ is θ̂ = (Σki)/n, where ki is the observed value for each sample and n is the sample size. 2. For the statistic c(Y1 + 2Y2) to be an unbiased estimator of θ1, c must equal 0.5. 3. W2 can only be an unbiased estimator of θ2 if the variance of W is 0, meaning W would have to be a constant value.

Uploaded by

Ashok Thiruvengadam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

IE 6200 Solutions: Homework 8 November 14, 2021

Name

1. Suppose a random sample of size n is drawn from the probability model


2
θ2k e−θ
PX (k; θ) = , k = 0, 1, 2, . . .
k!

Find a formula for MLE, θ̂.


Solution:
n 2 2
Y θ2ki eθ θ2(k1 +k2 +···+kn ) e−nθ
L(θ) = = Qn
i=1
ki ! i=1 ki !
Xn n
Y 
l(θ) = ln L(θ) = 2( ki ) (ln θ) − nθ2 + ln ki !
i=1 i=1
n
! 
dl(θ) X 1
= 2 ki − 2nθ
dθ i=1
θ
n
! 
dl(θ) X 1
= 0 → 2 ki − 2nθ̂ or
dθ θ=θ̂ i=1 θ̂
v
uX n
ki
u
u
t
i=1
θ̂ =
n

2. A random sample of size 2, Y1 and Y2 , is drawn from the pdf


1
fY (y; θ) = 2yθ2 , 0<y<
θ
What must c equal if the statistic c(Y1 + 2Y2 ) is to be an unbiased estimator for θ1 ?

Solution:
Z 1
θ 2 1
E[Y ] = 2 y 2 θ2 dy =
0 3 θ
 
2 1 4 1 2c
E[c(Y1 + 2Y2 )] = c[E[Y1 ] + 2E[Y2 ]] = c + = .
3 θ 3 θ θ

For the estimator to be unbiased, 2c = 1 or c = 0.5.


3. Suppose that W is an unbiased estimator for θ. Can W 2 be an unbiased estimator for θ2 ?
Solution:
E(W 2 ) = Var(W ) + E(W )2 = Var(W ) + θ2 . Thus, W 2 is unbiased only if Var(W ) = 0, which in essence means
that W is constant.
4. An electrical firm manufactures light bulbs that are known to have a population standard deviation of 40 hours.
A sample of 49 bulbs are found to have an average life span of 780 hours

(a) Find a 97% confidence interval for the population mean.


Solution: Known Population variance σ 2 so use standard normal distribution (Z)

σ = 40, n = 49, x̄ = 450, 1 − α = 0.97 ⇒ α = 0.03

2-sided confidence interval α2 = 0.015 ⇒ z0.015 = the critical value leaving an area of 0.015 to the right.
But the Z-table lists areas to the left of critical values.
Looking for 1 − 0.015 = 0.985 ⇒ P (Z < 2.17) = 0.985 ⇒ z0.015 = 2.17. or
Looking for 0.015 ⇒ P (Z < −2.17) = 0.015 use symmetry z α2 = −z α2 = −(−2.17) = 2.17.
σ σ
x̄ − z α2 √ < µ < x̄ + z α2 √
n n
40 40
780 − 2.17 √ < µ < 780 + 2.17 √
49 49
780 − 12.4 < µ < 780 + 12.4
767.8 < µ < 792.4

(b) If you want a narrower 97% confidence interval than the one found in the part (a), what can you do?
Solution: Increase sample size n.
(c) Find a 99% confidence lower bound for the population mean.
Note: If you can’t find the exact entry you are looking for in the table, then use the closest
one.
Solution: 1 − α = 0.99 ⇒ α = 0.01

P (Z < 2.33) = 0.9901(closest entry to 0.99), z0.01 = 2.33


σ 40
µ > x̄ − z0.01 √ = 780 − 2.33 √
n 49
µ > 766.69.

5. As an engineer working for a electrical power company you need to design an experiment to determine a 95%
confidence interval for µ the mean daily power usage for homes in Salt Lake City. Measuring the daily power
usage of every single home is not practical; therefore you need to choose a sample. What size does your sample
need to be to ensure that the confidence interval for is not larger than ±5kWh? Or in other words, what is the
minimum sample size that can be used to be 95% confident that the error between the sample mean x̄ and the
population mean µ will not exceed 5 kWh? Assume that the standard deviation for the population distribution
of power usage is known to be σ = 50kWh.
Solution: 2
 z α σ 2 
1.96 50
z α2 = z0.025 = 1.96, n = 2
= = 1536.64 ≈ 1537
d 2.5
k
6. If (0.57, 0.63) is a 50% confidence interval for p, what does n equal and how many observations were taken?
Solution:
v   
u k k
u
u (1 −
k t n n
− 0.67 = 0.57
n v  n 
u k k
u
u (1 −
k t n n
+ 0.67 = 0.63
n n
Adding the two equations gives 2 nk = 1.20 or k
n = 0.60
r
k (0.60)(0.40)
Substituting the value for n into the first equation above gives 0.60 − = 0.57. Solving this equation
n
for n gives n = 120.

7. Suppose that p is to be estimated by X


n and we are willing to assume that the true p will not be greater than 0.4.
What is the smallest n for which Xn will have a 99% probability of being within 0.05 of p?
Solution:
2
z0.005 p̂(1 − p̂) 2.582 (0.40)(0.60)
Take n to be smallest integer ≥ 2
= = 639.01 = 640.
(0.05) (0.05)2

You might also like