You are on page 1of 4

# NAME

:

ID:

University of Waterloo
Department of Statistics and Actuarial Science
STAT 330 Term Exam 1
October 10, 2012
• Answer all questions. The number in brackets [··] denotes the mark assigned to each question.
• Write your solutions in the space provided on the question sheets. The exam consists of 4
pages.
• Print your name and UWaterloo ID on each page in the space provided.
Q1: 20

Q2: 40

Q3: 20

Q4: 20

Total: 100

Question 1 [20 points]
Let f (x) = 2x, 0 < x < 1, zero elsewhere, be the probability density function (pdf) of X.
a. Compute E(1/X) using the pdf of X.
Z
E(1/X) =
0

1

1
f (x)dx =
x

Z
0

1

1
( )(2x)dx = 2.
x

b. Find the cumulative distribution function (cdf) and the pdf of Y = 1/X.
For y > 1,
Z

1
1
Py (Y ≤ y) = Px ( ≤ y) = Px (X ≥ ) =
X
y
fy (y) =

1

1 .

(2x)dx = x .

2.

dy y c. zero elsewhere. Compute E(Y ) using the pdf of Y and compare this result with the answer obtained in part a. Z E(Y ) = ∞ Z yfy (y)dy = 1 1 ∞ . 1/y = 1−(1/y)2 . y > 1. y > 1. zero elsewhere. 1/y d 2 Py (Y ≤ y) = 3 .

∞ 2 1 .

.

y( 3 )dy = (−2) .

y y 1 They are the same. = (−2)(−1) = 2. 1 .

Z 1 Z 1 1 Z 1 Z c(x − y)dxdy f (x. y) = c(x − y). Find c. a. y)dxdy = 0 y 0 .NAME: ID: Question 2 [40 points] Suppose X and Y have the joint pdf f (x. 0 ≤ y ≤ x ≤ 1.

1  1 2 .

.

x − y(1 − y) dy c 2 .

y 0  Z 1 1 2 c (1 − y ) − y(1 − y) dy 2 0  Z 1 1 1 2 c + y − y dy 2 2 0 .

1 .

1   1 1 3 .

.

1 2 .

.

c + y − y 2 6 .

0 2 .

y) : 0 ≤ y ≤ x ≤ 1}. P (XY ≤ 1) = 1 because XY ≤ 1 for every (x. c. 6 Z = = = = = (1) y 1  (2) (3) (4) (5) (6) b.e. c = 6. 1 4 Z P (X + Y ≤ 1/2) = 1 −y 2 Z 0 Z 1 4  = 6 0 = = = 1 4 . i. Find P (X + Y ≤ 1/2). y) ∈ A = {(x. Find P (XY ≤ 1).0 c = 1.

1 −y  1 2 .

.

2 1 x − y( − y − y) dy 2 .

y 2  1 1 1 2 2 2 6 {( − y) − y } − ( y − 2y ) dy 2 2 2 0  Z 1 4 1 6 − y + 2y 2 dy 8 0 .

1 .

1   1 1 1 2 .

.

4 2 3 .

.

4 6 ( )( ) − y .

+ y .

8 4 2 0 3 0 1/16. Z = 6(x − y)dxdy (7) y (8)  (9) (10) (11) (12) 2 .

NAME: ID: d. . Find the marginal pdf’s of X and Y .

x Z x .

2 2.

f1 (x) = 6(x − y)dy = 6x − 3y .

0 < x ≤ 1. 0 Z 1 f2 (y) = y 0 . = 3x2 .

1 .

2.

6(x − y)dx = 3x .

Un independently from a uniform(0. β > 0. W β ) ≤ x] (13) α W P[ ≤ x1/β ] (14) α P [W ≤ αx1/β ] (15) 1/β −x F (αx ) = 1 − e . y) 2(x − y) = . (2) The joint pdf can not be factorized by non-negative functions g(x) and h(y). 0 ≤ y < 1. (17) dx variables be generated from a uniform(0. Find the pdf of X. f1 (x|y) = No. 0 ≤ y < 1. f (x. y) ∈ A). y) = . If W follows a Weibull distribution. Therefore. For x ≥ 0. y e. Are X and Y independent? Briefly explain in 3 different ways. The random variables Xi = F (Ui ) = α − log (1 − Ui ) for i = 1. (16) d P (X ≤ x) = e−x . 3 . y) ∈ A (or f2 (y) 6= f2 (y|x) for every (x. How could Weibull random ator? Notice that F (X) follows Unif(0. Question 3 [20 points] The Weibull cdf is x β F (x) = 1 − e−( α ) . − 6y(1 − y) = 3(y − 1)2 . · · · . · · · . n are independent Weibull random variables. y ≤ x ≤ 1. 0 < x ≤ 1. a. where U ∼ Unif(0. Let X = (W/α)β . α > 0. 2. This is because (1) The support set A of the joint pdf is not a cross-product.1) random  1/β −1 number generator. zero elsewhere. for x ≥ 0. Find the conditional pdf’s of (X|Y = y) and (Y |X = x). zero elsewhere.1) distribution and F −1  (U ) = α − log (1 − U ) 1/β .1) random number gener- P (X ≤ x) = P [( = = = f (x) = b. U2 . X and Y are dependent. for x ≥ 0. if we can generate U1 . 0 ≤ y ≤ x. f2 (y) (y − 1)2 2(x − y) f (x. (3) f1 (x) 6= f1 (x|y) for every (x. f2 (y|x) = f1 (x) x2 f.1). x ≥ 0.

Notice that (24)→(25) is because (25)→(26) is because when −h < t < 0. the function etx is decreasing in x. R x<a etx f (x)dx ≥ 0. 0 < t < h. − h < t < 0. for some h > 0. (22) That is. from (18) to (21) and 4 . Z ∞ M (t) = etx f (x)dx (23) −∞ Z = tx Z etx f (x)dx e f (x)dx + x>a Z ≥ (24) x≤a etx f (x)dx (25) eta f (x)dx (26) x≤a Z ≥ x≤a ta = e P (X ≤ a). P (X ≥ a) ≤ e−at M (t). For 0 < t < h. Z ∞ etx f (x)dx M (t) = (18) −∞ Z tx = Z etx f (x)dx e f (x)dx + x≥a Z ≥ (19) x<a etx f (x)dx (20) eta f (x)dx (21) x≥a Z ≥ x≥a ta = e P (X ≥ a). For −h < t < 0. − h < t < h. Proof. 0 < t < h and that P (X ≤ a) ≤ e−at M (t). If X is a continuous random variable with pdf f (x). the function etx is increasing in x. let from (23) to (26). If X is a discrete random variable with pmf f (x). (27) That is. P (X ≤ a) ≤ e−at M (t). Notice that (19)→(20) is because (20)→(21) is because when 0 < t < h. P substitute for R R x>a etx f (x)dx ≥ 0. Prove that P (X ≥ a) ≤ e−at M (t). −h < t < 0.NAME: ID: Question 4 [20 points] Let X be a random variable with moment generating function (mgf) M (t).