f I

1
I
1
I
1
·r r r 1 '

'
I r

I~f~J1f~o~M 11111)
T E C H N IS C H E if 'i

Lecture·• Sta t·1St·JCS

Stu de nt ID:

Se at number:

Semester:
wsxx
Lecturer: This is jus t a tes t exam!
Prof. Dr. Markus Mayer
.
.

Qu est ion 1 (13 Points): their computers last we

ek.
ed 6 stu den ts ho w ma ny times they rebooted
You ask
rs and 2 PC users.
• There were 4 Mac use
2 an d 3 times.
• The PC users rebooted
1, 2, 2 an d 8 times.
• The Mac users rebooted de nt (M ac =
the typ e of com pu ter of a randomly chosen stu
dom variable representing
Let C be a Bernoulli ran
0, PC = l). es values 1, 2, 3, 8).
a ran do ml y cho sen stu dent rebooted (so R tak
Let R be the nu mb er of
times ba bil ity ma ss
C an d R. Be sur e to include the marginal pro
joi nt probability table for
(a) (6 Points) Create a
functions.
E(C) and E(R).
(b) (3 Points) Compute
on the nex t page!
The question continues

Q )

0
·1 0 1.: :i-
r'
V
, f
---;-
l

ö
-
f

b 6

k;J lc ~ ) 3
I (r )

Page 2 of 12
 11il)
D
-

1
Stud ent ID:
TE CH NI SC HE
HOCHSCHULE Sea t number:
DEGGENDORF
Lecture: Statistics Semester: / wsxx
Lecturer: Prof. Dr. Markus Mayer This is just a test exam!

c) JP {c .: ,P -j )( r, -f c) (I J (( )
I( k) :: ]
f( () = j

i C1 - 3)u- 3 ) -+ iC j g)( 3- 3) + ; -( {j - n{ 1- J ) }(r / ; I +;(,-; )( n)

:::-(/ • ;) +- (f }{ f) -f:l)+ -{ ; {-:/}) 1-· J (- J) 5

/1
G re, '
p{Pe -rt.- <Y'o bS
1tt?j o,,,f. :(,J(. .. C'o' o/,/,,.,,..,
[ a , cl < /, c, l~

CJH,,u--- f t1 U? - 7 c-nd ( at,-e_

C and Ran d explain its significance

for how C and Rar e related.
ts) Dete rmin e the covariance of
(c) (4 Poin t?
's called for.) Are Ran d C independen
(A one sentence explanation is all that

<-tj

-~---
r{? {J ( C rR ~------,-:1
_ _ _ _ __

Page 3 of 12
1
 1 TE CH NI SC HE
HO CH SC HU LE
DE GG EN DO RF
ß1111) Stud ent ID:
Sea t number:

Semester: wsxx
Lecture: Stat istic s

Prof. Dr. Markus Mayer This is just a test exam!

Lecturer:

Que stio n 2 (6 Points):

renheit and degrees Celsius
Recall the relation between degrees Fah

degreesCelsius = i·
degreesFah.renheit - i;o
mer in Los Angeles and San
ture in degrees Fahrenheit for the sum
Let X and Y be the daily high tempera
tures in degrees Celsius.
Diego. Let T and S be the same tempera r(T, S) (Corr is Correla-
that Cov (X, Y) = 4 and Cor r(X , Y) = 0.8. Compute Cov(T, S) and Cor
Suppose
tion).
-46 0
F C,

:: ln r { )<+b (
c c><, '1 1 -1
C 7'-f p(_ )
,
c
- q_ . cir - J ;:::

Co<vr(7 1·f }= 0 18 ::: Co-r1r (y ( yl

>,.,.."-t- et.. , : , . A r tJ- , tt-. . t<..

.sc,,_

Page 4 of 12
 ,,

11iil)
if
:T
-
I
1 TE C H N I SC H E
HOC HSC HUL E
DEG GEN DOR F
Lectu re: Statis tics
Stude nt ID:
Seat number:

Seme ster: 1 wsxx

I
Lectu rer: Prof. Dr. Markus Mayer This is just a test exam!
I

1\
Ques tion 3 (15 Points):
mastered 70% of the material. Assume this means
You are taking a multiple choice test for which you have
E m test question, and that if you don't know the
that you have a_0.7 chance of knowing the answer to a rando
four answer choices. Finally, assume that this holds
answe r to a question then you randomly select among the
for each question, independent of the others.
on the exam?
(a) (5 Points) What is your expected score (as a percent)
t. You likely found p in part (a), but in any case
Let p be the probability of getting a random question correc
you can just use the letter p for this p~obability.
you should assume 0. 7 < p < 0.9. In parts (b), (c) and (d)
bility you score 90% or higher? (Do not simplify
(b) (3 Points) If the test has 10 questions, what is the proba
the expression you get.)
The question continues on the next page!

c_

f-= 0. 7=?§

-7=7, 5 -~

Page 5 of 12
 11 il)
lfi
r
I TEC H NISC HE
HOC HSCH ULE
DEGG ENDO RF
Lecture: Statisti cs
Student ID:
Seat number:
Semest er: wsxx
Lecturer: Prof. Dr. Markus Mayer This is just a test exam!

correct? (Again, do not simplify

(c) (2 Points) What is the probability you get the first 6 questions on the exam
the express ion you get.)
rather have a test with 10 or 100
(d) (5 Points) Suppose you need a 90% to keep your scholarship. Would you
questio ns? Why? (Your answer only needs tobe one sentenc e long.)

d,) ~s p ~s.'-
0. 7-L:... P.c: o.rr ~e ~au/

..!-tJni i~ (3g - , ~
- ;t

,/4,t,-'?- fi4-V 46~

/ l tPf c~ bt2- bYj/IP d-ed

loe 4'/4 54 .i~> /45s ,6-t.~;r_ ;ö I ~ a...pt.ef

400 sz9-
p~-c~ fCJ

6
c) o. 7- ::r:5 ö,'2. ~

Page 6 of 12
 1 TECHNISCHE
HOCHSCHULE
DEGGENDORF
ii11)
11 Student ID:

Seat number:

Lecture: Statistics Semester: wsxx

Lecturer: Prof. Dr. Markus Mayer This is just a test exam!

Question 4 (10 Points):

= kx 2• Let Y = X
3
Let X have range (0, 3] and density fx(x) •

(a) (4 Points) Find k and the cumulative distribution function of X.

(b) (3 Points) Find the 30th percentile of X.
(c) (3 Points) Compute E(Y).

Page 7 of 12
 l
I TECHNIS CHE1fi
HOCHSCHU LE
DEGGENDO RF
Lecture:
1111)
Statistics
Student JD:
Seat number:

Semester: wsxx
Lecturer: Prof. Dr. Markus Mayer This is just a test exam!
. . VUA/("J,,..(
- •
Question 5 (14 Points): .,1l
: u _set up a 2D CCD sensor (a 2D array of pixel cells) in your lab with 100 byl100 pixels. The CCD is ideal,
0

l.e. 1t ~oes not need calibration and the single pixel measuremen Xi are independent from each other over
multiple captures. You want to test your sensor. Therefore you e ose the CCD to a uniform (over all pixels)
light source and thus all pixel measurements Xi can be seen a i.i.d random variable values over the sensor
area and over multiple captures. You know that these values are Poison distributed.
You need speci.fic no knowledge about the Poison distribution to answer the following questions.
(a) (5 Points) With theRrm light source you capture an irnage. You compute the average µ of all the pixel
2
values Xi and this i ~ You compute the average of all (xi - µ) and this is also 100. Then you teil your
colleague: "For my further computations, I assume that the expected value of the Poison distribution is 100
and its variance is also 100 at my illumination strength." Is this justified? Explain in a few sentences.
The question continues on the next page!

./(oo x · -foo ('·) e-c s _"":

r= --1 (Y0

Vtuz-
1~ t-ci
(x~-?)2 - -t:5 /2i/
2
L,OL ;{/ -/ A11a{r-:xi-/.,.__l J

/4 ~, l'4L (Uul-r.... d,t,

1/ v..e.

Page 8 of 12
 I TE CH NIS CH Efi
HOC HSC HUL E
DEG GEN DOR F
Lecture:
l li l)
Statis tics
Stude nt 1D:
Seat number:

Semester: 1 wsx x

Lecturer: Prof. Dr. Markus Mayer This is just a test exam!

What is the rough probability that the sum over all

(b) (5 Points) Let's just say your assumption is justified.
the pixels on your image is in range [998000; 1002000]?
by averaging multiple image captures. "Average"
(c) (4 Points) You want to decrease the measurement noise
for each pixel position on the CCD separately.
in this case means averaging the values of multiple captures
t noise of 1 at the illumination strength of
What you'll accept is a standard deviation of the measuremen
ed image. How many images do you have to
your unifo rm light source on the pixel values of the averag
average at least to achieve that?
-1oox '1t>o f,yerr;
er..=. -fo
l-"? AJ .::. ,{"'~C')

1- }J . <!r l}_ f-~ - Jt • JA

- 1~c:io~ooo; :::. Aa:.o
ACD 2~0 -f'-5

cr-s
PC oo ~)
1__;!.o 2.oo - 4
-(0~

)
p (
? (
--- --
-~c:, ,-(."J
--,or., o

- 2. <. t- <
2)
./
~('_;)
-4~,...,c--.,

--
.,,,......_. o. o/5

-- r
/J
-1
---
)(1
p lJ • X
j-
C ) , tl j '=-1 (

Vc-~{ >< )= 1~ AJ'2,.

·)J . Vo.1/'{ )C--i_) - - VtJ,-(ll.:I.}_
Al

IJ =- AO O
-

Page 9 of 12

L
 r
l
I I IJ7 I)
lt

TE CH NI SC HE li Stud ent ID:

HO CH SC HU LE Seat number:
DE GG EN DO RF
Lecture: Semester: WSX X
Stati stics

Lecturer: Prof. Dr. Markus Mayer Thls is just a test exam!

• • •
Que stion 6 (9 Points):
The Pare to distribution with param eter a has range
(1, oo) and pdf
Cl'
f(x) = -xcr

Supp ose the data

5,2,3
of a.
n. Find the maximum likelihood estimate (MLE)
was draw n independently from such a distributio

~s
i { 5,2-, 3 / o<.) = s.;_ -2~ -3 ,
(>(._

o(._ -
- -
3o
O(__

t.Y<-~~tfo
lo L,7 1e, ,& ,-~ So
J ec._ > CL '3 -_;ltA 3o
l"' { f { 5, 2, ) / ex..) ) =-- --l,,fA- - 3oot- =-
~b t C<.-
3 - Lvt t's aJ
·':..::

. . I
~h { f { cJ_ cvtt { / c;,<-) }

Page 10 of 12
 ______________
I
.._,

I~~cf
TE CH NIS CH EJ i

Lecture·
1111)
Statis tics
Stude nt 1D:

Seat number:

Semester: wsxx
•
Lecturer: . Prof. Dr. Markus Mayer This is just a test exam!

• • •
Que5tion 7 (14 Points): •
ed
ining two 4-sided dice, one 8-sided die, and one 12-sid
Y~ur friend grabs a die at random from a drawer conta
is 5.
die. She rolls the die once and reports that the result
the prior, likelihood, and posterior for the type of die
(a) (7 Points) Make a discrete Bayes table showing
rolled given the data.
of rolling a 5?
(b) (2 Points) What is the prior predictive probability
your probability that the next roll will be a 7?
(c) (5 Points) Given the data of the first roll, what is

II; ßo- t?~ 7ell/'1 ;a)~(

Pv-rt7v /, l!elt1,. &~~

-;i_ ()
(l,) /oft.R-67:5
((5 /lf) ()
(P -ß f -= _-1- 0

,~
2-_
(3! / '2.
--1
4-
-f
11
.A
8, 7; 25 6 0(6
@r /{
A -
1 .--- - -
{J 1 Z t: f 7/f1- tr
::- 5:/¤
.jul f '1
---,
0 - 316t:J"
o, tr
t::). ööS 6
ö, ~1 >

b.) <tJ) 5b
I?"')
60

j,'k efi t~ ß~ ~7.ep.~ {¾.,,'q,,.-

( f)/,,(cn-
c. ) r/01 0fh« 1-5

----i
0 I
(), 6'11/f-
----,
- _ö,6 1
0- 0 5?5 G, t Ö~'
e
-<---.:: :

O·J f
Ji -o.) f (!) . oo 0. -1-r g5

Page 11 of 12
! P· o. o6 r 5
 l l ~ § ; :SS: i:i§..__ ',
,/._L-.:;;..~

,
I TECHNISCHETI
HOCHSCHULE
DEGGENDORF
Lecture:
1111)
Statistics
Student ID:

Seat number:

Semester: wsxx
Lecturer: Prof. Dr. Markus Mayer This is just a test exam!

Question 8 (9 Points):
Everyone knows that giraffes are tall, but h.,....."'""__.,,.-uch do they weigh? Let' se that the weight of male
giraffes is normally distributed with mea 200kg d standard deviati
I volunteered .at the zoo and was given the s of weighing their e Beau. Now weighing a giraffe is
not easy and the process produces random errors following a ~istribution. To compensate for the
inaccuracy of the scale I weighed Beau three times and got the following measurements:

I250kg,I300kg,I350kg.

What is the posterior expected value ofBeau's weight?

,1 2 r; o -1
X - .-f 3C?ö -( ·13 'io_
Cl

/(

--

·-· fto~(

Page 12 of 12