the final exam solutions. part i, #1, central limit theorem let x1,x2, …, xn be a sequence of...

12
The final exam solutions

Upload: charles-bridges

Post on 17-Jan-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

The final exam solutions

Page 2: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part I, #1, Central limit theorem

Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance σ2

The sum of a large number of independent random variables has a distribution that is approximately normal

variablerandom normal standard aely approximat is ...

varianceand mean with normalely approximat is ...

21

221

n

nXXX

nnXXX

n

n

Page 3: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part II #2 If the successful probability of trial is very small, then

the accumulatively many successful trials distribute as a Poisson.

The Poisson parameter λ is the mean frequency of interest within a large bundle of experiment

The exponential distribution is often used to describe the distribution of the amount of time until the specific event first occurs.

The exponential distribution seems to partition the Poisson distribution with a parameter λ by every two occurrences into several small time intervals and focuses on a specific time interval.

The same distributional parameter λ, The exponential distribution emphasizes the cycle time, 1/λ,

while the Poisson on the frequency λ.

Page 4: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part I, #3

n

ZZZ

nX

ZT n

n

n

222

21

2n

2

...

n

X ,

/

If Z and Xn2 are independent random variables, with Z having a std. normal dist. And Xn2 having a chi-square dist. with n degrees of freedom,

•For large n, the t distribution approximates to the standard normal distribution

Page 5: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part I, #4

When these two population variances are unknown but equal, we calculate the pooled variance estimator, Sp2, by means of weighted average of individual sample variance, S12 and S22

Page 6: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part I, #5 It is not correct. He had better say that the statistic or random

variables used to obtain this confidence interval is such that 95% of the time that it is employed it will result in an interval covered the μ.

He can assert with probability 0.95 the interval will cover the μ only before the sample data are observed.

Whereas after the sample data are observed and computed the interval, he can only assert that the resultant interval indeed contains μ with confidence 0.95.

Page 7: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part I, #6

)0log

or(,0 means )/x,,x,(x Max

values.observed offunction likelihood ofy probabilit the

maximize would,by denoted , ofestimator likelihood maximum The

found. be

also could of value true theif )x,,x,(x sample observed by the

determined be will)/x,,x,(xfunction likelihood of valueThe

sample. random a dconstitute and

observed be toare ,parameter unknown an for except given assumed is

on distributijoint whose,X ,X ,X variablesrandom i.i.d. theSuppose

n21

^

n21

n21

n21

fd

dff

f

Page 8: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part I, #7

(a) if d(X) is the estimator of θ, and E[d(X)]=θ, then the d(x) is unbiased.

(b) the MLE estimator is not unbiased.

(c) the sample variance is an unbiased estimator of σ2

n

Xxn

ii

2

1^

2

)(

1

)( 2

12

n

XxS

n

ii

Page 9: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part II #1

Mean=np=90, Var=np(1-p)=150(.6)(.4)=36, ∴stand

ard deviation=6 P{X≦80}=p{X<80.5}=P{(X-90)/6<(80.

5-90)/6}=P{Z<-1.583} =1-P{Z≦1.583} (=1-0.943=0.057)

Page 10: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part II #2,3

unknown is en thesmaller wh is intervalupper confidence 95% of boundlower The

confidence 95% with )(86330.45, ) ),4/9400(735.190450(say we

},/{,1}/

)({

,~/

)( unknown, is σ if

confidence 95% with )86584.25,()),4/9400(645.190450(

},/{,1}/

)({

),1,0(~/

)( known, is σ if

1,

__

1,

1

_

__

_

nStXPnS

XtP

tnS

X

nzXPn

XzP

Zn

X

nn

n

Page 11: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part II #4

confidence 95% with )244.0 ,096.0(

95.0100/)83)(.17(.96.117.100/)83)(.17.0(96.117.

1/1/1

ion)approximatnormality for enough large isn that assume (we ),1,0(~

1

83),100(.17)(. varianceand 17mean on with distributi binomial a

obey toseem itemsproduct 100 theSo p. is rate deffective theSuppose

2/2/

p

pP

n)p-(pzppn)p-(pzpP

N

)p-(pn

pX-n

^^^^^^

^^

^

Page 12: The final exam solutions. Part I, #1, Central limit theorem Let X1,X2, …, Xn be a sequence of i.i.d. random variables each having mean μ and variance

Part II #5,6

decreasinglightly %,75.20274.01068

)7.0)(3.0(96.1error ofmargin the

0.3 isresult servey theif

size samplenecessary theis 1068 ,11.1067

)03.0(

)5.0)(5.0(96.1,03.0

)5.0)(5.0(2

2

25.0

n

nn

z