Copyright reserved

Faculty of Natural and Agricultural Sciences

Department of Statistics
Mathematical Statistics (WST 211)
Examination: Memorandum 9 June 2015

Internal Examiners: Dr. R. Ehlers

Mr. J. Ferreira
External Moderator: Dr. I. Fabris-Rotelli
Time: 3 hours
Questions: 12 questions + 15 multiple choice questions
Pages: 11
Marks: 100
Perusal: 19 June 2015; 9:30-11:30; Venue ???
Supplementary exam: 26 June 2015; 9:30-12:00; Venue ???

INSTRUCTIONS:
• Answer:

— Questions 1 to 6 in Book 1 (30 marks)

— Questions 7 to 12 in Book 2 (40 marks)
— Answer the multiple choice questions on SIDE 1 of the multiple choice answer sheet
(30 marks)

• Use the correct notation. No marks will be awarded without the necessary explanations.

• List of formula is attached.

• Work to 4 decimal places.

 1

1. Suppose that fX is continuous in the vicinity of x. We then have that

1
{FX (x + ∆x) − FX (x)}
∆x

1 x+∆x
 x
= fX (t)dt − fX (t)dt by Definition
∆x −∞

x+∆x  −∞
1 
= fX (t)dt
∆x x
1
= × {∆x × fX (θ)} for ∆x small enough and some θ between x and x + ∆x
∆x
by the mean value theorem for integration since fX is
continuous in the vicinity of x
= fX (θ) and hence
 1 

FX (x) = lim ∆x {FX (x + ∆x) − FX (x)} by definition of FX (x)
∆x→0
= lim fX (θ)
∆x→0
= fX (x) since θ → x as ∆x → 0 and fX is continuous at x.

2. Let fW and FW be the density and distribution functions respectively of W .

Then P [W > t] = P [no event i.e. 0 events take place in [0, t]]
= e−λt (λt)0 /0!
= e−λt i.e.
FW (t) = P [W ≤ t] = 1 − P [W > t] = 1 − e−λt provided t > 0 and
FW (t) = P [W ≤ t] = 0 if t < 0 since waiting time is > 0

 0 for t < 0
d
Therefore fW (t) = dt
FW (t) =
 λe−λt for t > 0
1
i.e. W is an exponential random variable with parameter .
λ

3. FX2 (x2 ) = P (X2 ≤ x2 )

= P (X1 < ∞, X2 ≤ x2 )
∞ x2
= fX1 ,X2 (t1 , t2 ) dt2 dt1
−∞ −∞
x2 ∞
= fX1 ,X2 (t1 , t2 ) dt1 dt2 since we can change order of integration
−∞ −∞
x2
= fX2 (t2 ) dt2
−∞
∞
where fX2 (x2 ) = fX1 ,X2 (x1 , x2 ) dx1 for all x2 ; ∴ it is the pdf of X2 .
−∞
 2

4. MY (t) = E[etY ] = E[et(X1 +X2 +...+Xn ) ] = E[etX1 .etX2 . . . etXn ]

   
= E[etX1 ].E etX2 . . . E etXn since Xi s are independent
= MX1 (t) MX2 (t) . . . MXn (t)

5.

(a) P (X > 0.5) = 1.

∞
(b) E (X 2 ) = x2 (2.5) (0.6)2.5 x−3.5 dx
0.6
(2.5)(0.6)2.5 −0.5 ∞
= −0.5
x |0.6

= 1.8
m
(c) (2.5) (0.6)2.5 x−3.5 dx = 0.5
0.6
(2.5)(0.6)2.5 x−2.5 m
−2.5
|0.6 = − (0.6)2.5 m−2.5 + 1 = 0.5
∴ m = 0.7917.

6.
1
(a) For 0 < x < 1 : fX (x) = x
2 (x + y) dy

= (2xy + y 2 ) |1x

= 2x + 1 − 3x2

fX,Y (x,y)
For 0 < x < 1 : fY |X (y|x) = fX (x)

2(x+y)
= 2x+1−3x2
for x < y < 1 and 0 otherwise.
 0.5 2(x+y)  0.5 0.6+2y
(b) P (Y < 0.5|X = 0.3) = 0.3 2x+1−3x 2 dy = 0.3 1.33
dy = 0.2105
(c) FX,Y (x, y) for 0 < x < 1 and y ≥ 1 :
 y t
FX,Y (x, y) = 0 0 2 2 (t1 + t2 ) dt1 dt2
y 
= 0 (t21 + 2t1 t2 ) |t02 dt1
y
= 0 3t22 dt1

= t32 |y0

= y3
 3

  ∞
7. EX1 EX2 |X1 (X2 |X1 ) = EX2 |X1 (X2 |x1 ) fX1 (x1 ) dx1
−∞
∞ ∞
= x2 fX2 |X1 (x2 |x1 ) dx2 fX1 (x1 ) dx1
−∞ −∞
∞ ∞ fX1 ,X2 (x1 , x2 )
= x2 fX1 (x1 ) dx2 dx1
−∞−∞ fX1 (x1 )
∞ ∞
= x2 fX1 ,X2 (x1 , x2 ) dx2 dx1
−∞−∞

= E (X2 ).

8.

(a) S = {O1 O2 O3 . . . O8 | Oi = E1 or E2 or E3 or E4 ; i = 1, 2, ..., 8}.

8!
(b) = 1680.
1!2!3!2!
 
(c) E (X2 X4 ) = EX2 EX4 |X2 (X2 X4 |X2 )
 
= EX2 X2 EX4 |X2 (X4 |X2 )
= EX2 [X2 (8 − X2 ) 0.5]
= EX2 (4X2 − 0.5X22 )
= 4 (8) (0.2) − 0.5 (3.84)
= 4.48

9.

P (X=1,Y ≥1) 0.30

(a) P (X = 1|Y ≥ 1) = P (Y ≥1)
= 0.75
= 0.4.

(b) E (XY ) = xyfX,Y (x, y)

∀x ∀y

= 0.1 + 2 (0.2) + 2.5 (0.15) + 2.5 (2) (0.3)

= 2.375

(c) E (X 2 |Y = 1) = x2 fX|Y (x|y = 1)

∀x
fX,Y (x,1)
= x2 fY (1)
∀x
fX,Y (1,1) f (2.5,1)
= (1)2 0.25
+ (2.5)2 X,Y0.25
= (1)2 0.10
0.25
+ (2.5)2 0.15
0.25

= 4.15
 4


 0 for y ≤ 0
10. FY1 (y) =
 1 − (1 + y)−n for y > 0.

Gn (y) = P (Zn ≤ y)
= P (nY1 ≤ y)
= P (Y1 ≤ y/n)
= FY1 (y/n)

 0 for y ≤ 0
=
 1 − (1 + y/n)−n for y > 0

 0 for y≤0
i.e. lim Gn (y) =
n→∞  1 − e−y for y>0 from Theorem 15
i.e. Zn converges in distribution to an exponential distribution with parameter 1.

11.


 0 for x ≤ 0


(a) Graphs of FX (x) = x3 for 0 < x < 1 and



 1 for x ≥ 1.


 0 for x < 0.27




 0.2 for 0.27 ≤ x < 0.68


F5 (x) = 0.6 for 0.68 ≤ x < 0.75





 0.8 for 0.75 ≤ x < 0.83



1 for x ≥ 0.83.
(b) A value was generated randomly from U NIF (0, 1) distribution and set equal to FX (x) .
u = FX (x) = x3
1
∴ x = u3
∴For x = 0.68, u = x3 = 0.3144.
 5

12.
√ √
(a) For y = (x − 3)2 there are two inverses: x = 3 − y and x = 3 + y.
k  
 d −1 
Use Theorem 11 to determine fY (y) : fY (y) = fX (u−1
j (y))  u
dy j
(y) .
j=1
√ √
For 0 < y < 1 there are two inverses: x = 3 − y and x = 3 + y.
√
For 1 < y < 4 there is only one
  inverse: x = 3 − y  
√  √   √ d  √ 

 f (3 − y)  d
3 − y  + f (3 + y)  3 + y  for 0 < y < 1
 X
  dy   
X dy
√ d √ 
fY (y) = fX (3 − y)  dy 3− y  for 1 < y < 4



 0 otherwise

 1 √ √

 (3 − y)2 2√1 y + 21 1
(3 + y)2 2√1 y for 0 < y < 1
 21
1 √
= 21
(3 − y)2 2√1 y for 1 < y < 4



 0 otherwise
(b)

i. fX1 ,X2 (x1 , x2 ) = fX (x1 ) fX (x2 ) since X1 , X2 are i.i.d.


 1 x2 · 1 x2 for 1 < x1 < 4 and 1 < x2 < 4
21 1 21 2
=
 0 otherwise.

z1 = x1 + x2 z2 = x1 − x2

z1 + z2 = 2x1 z1 − z2 = 2x2
z1 +z2 z1 −z2
x1 = 2
x2 = 2
 
 1 1
 2
2
J ((x1 , x2 ) → (z1 , z2 )) =   = −1
1 1  2
 2
−2 
From Theorem 12:
fZ1 ,Z2 (z1 , z2 )
   z1 −z2   1 
= fX z1 +z 2
2
fX 2
− 
2
    
 1 2 1 z1 +z2 2 z1 −z2 2
 
for 1 < z1 +z2
< 4 and 1 < z1 −z2
<4
21 2 2 2 2 2
=
 0 otherwise.
 1 2  3  3
ii. P (Y1 < 3) =2 21 1
y 2 y 2 dy2 dy1
y2 1 2

a=1 b=3 c=1 d = y1

 6

Multiple choice questions:

1. E

2. A

3. E

4. B

5. D

6. C

7. C

8. E

9. A

10. B

11. D

12. D

13. A

14. E

15. C