STAT 410

Fall 2016

Review
random variables

discrete

continuous

probability mass function

p.m.f.

probability density function

p.d.f.

p( x ) = P ( X = x )

f( x )

0 p( x ) 1

f( x ) 0

p( x ) = 1

f (x ) d x

all x

=1

cumulative distribution function

c.d.f.
F( x ) = P( X x)
F( x ) =

p( y )

F( x ) =

y x

f (y) d y

expected value
E( X ) = X
discrete

continuous

If

x p ( x)

< ,

x p ( x)

E( X ) =

all x

x f ( x) d x

discrete

continuous

If

g ( x) p ( x)

< ,

all x

E(g(X ) ) =

g ( x) p ( x)

all x

< ,

all x

E( X ) =

x f ( x) d x

If

If

g ( x) f ( x) d x

< ,

E(g(X ) ) =

g ( x) f ( x) d x

variance
Var ( X ) = 2X = E ( [ X X ] 2 ) = E ( X 2 ) [ E ( X ) ] 2

discrete
Var ( X ) =

continuous

( x X ) p ( x)
2

( x X )

Var ( X ) =

all x

f ( x) dx

x 2 p ( x)

[E ( X ) ] 2

all x

= x 2 f ( x) dx [ E ( X ) ] 2

moment-generating function
MX( t ) = E( e t X )

discrete
MX(t ) =

continuous

e t x p( x )

MX(t ) =

t x f ( x ) dx

all x

Example 1:

p( x )

F( x )

1
2
3
4

0.2
0.4
0.3
0.1

0.2
0.6
0.9
1.0

F( x ) =

0
0.2
0.6
0.9
1

x <1
1 x < 2
2 x<3
3 x<4
x4

p( x )

x p( x )

1
2
3
4

0.2
0.4
0.3
0.1

0.2
0.8
0.9
0.4

E ( X ) = X = 2.3.

2.3

p( x )

x 2 p( x )

1
2
3
4

0.2
0.4
0.3
0.1

0.2
1.6
2.7
1.6

E ( X 2 ) = 6.1
Var ( X ) = 6.1 2.3 2 = 0.81

6.1

M X ( t ) = E ( e t X ) = 0.2 e t + 0.4 e 2 t + 0.3 e 3 t + 0.1 e 4 t

Example 2:
Let X be a continuous random variable
with the probability density function

a)

f ( x ) = k x 2,

0 < x < 1,

f ( x ) = 0,

otherwise.

What must the value of k be so that

f ( x ) is a probability density function?

1)

f (x) 0,

2)

f ( x ) dx = 1 .

f ( x ) dx

kx

b)

dx

k x 2 dx

x3 1

0
3

1
k

k
3

k = 3.

Find the cumulative distribution function F ( x ) = P ( X x ).

f X ( x ) = 3 x

0 < x <1

x<0

F X ( x ) = 0.

0x<1

FX( x ) =

o.w.

3y

x1
fX( x )

c)

FX( x )

Find the probability P( 0.4 X 0.8 ).

P( 0.4 X 0.8 ) =

0. 8

f (x ) dx

0. 4
0.8

3 x

0.4

dx = x 3

0.8
0.4

= 0.8 3 0.4 3 = 0.448.

F X ( x ) = 1.

d y = x 3.

OR
P( 0.4 X 0.8 ) = F X ( 0.8 ) F X ( 0.4- ) = 0.8 3 0.4 3 = 0.448.

d)

Find the median of the distribution of X.

m
Need m = ? such that

( Area to the left of m ) =

f ( x ) dx =

1
=
2

m=

e)

f ( x ) dx =

3 x

dx = x 3

m
0

1
.
2

= m3.

= 0.7937.

Find X = E ( X ).

E ( X ) = X =

x f ( x ) dx = x ( 3 x

)dx = 3 x 3 dx .
1

x4 1
3

= 3
= 0.75.
=
4 0
4

f)

Find X = SD ( X ).

Var ( X ) = X =
2

x 2 f ( x) dx ( ) 2 =
X

1
3 x 4 dx 3
4

x 5 1 3 2
3
3
9

= 3
= 0.0375.
=

=
5 0 4
80
5
16

X = SD ( X ) =

Var( X ) =

0.0375 = 0.19365.

g)

Find the moment-generating function of X, M X ( t ).

MX( t ) = E( e t X ) =

e t x f (x ) dx =

dv = e t x dx,

du = 6 x dx,

v =

MX(t ) =

t x 3 x 2 dx .

e t x.

1
1
e t x 3 x 2 dx = 3 x 2 e t x
t
0

3
= et

1 t x
6 x dx
e

1 t x
6 x dx

t e
0

u = 6 x,

dv = =

du = 6 dx,

v =

t
3
MX(t ) = e t

e
0

u = 3 x 2,

e t x dx,

e t x.

1 1
3 t
1
1 t x

t
x
6 x dx = e 6 x e
e
t
t

t2

0 0
0

t 2 e

t x 6 dx

6
1
3 t
6
3
6
6
6
e 2 e t + 3 e t x = e t 2 e t + 3 e t 3 ,
t
t
t
t
t
t
t
0

t 0.

M X ( 0 ) = 1.

h)

Find E (

X ) and E ( ln X ).
1

E(

X) =

x 3x 2 d x =

6
.
7

E ( ln X ) =

ln x 3 x

1
dx = .
3

Example 3:
6

f X ( x ) = 5 x

x<1

F X ( x ) = 0.

x1

FX( x ) =

x >1

o.w.

5y

dy

= y 5

fX( x )

x
1

= 1 x 5.

FX( x )

6
5
x 5x d x = 5x d x =

E( X ) = X =

E( X ) =

5x

Var ( X )

5
= 1.25.
4

5
d x = 5 x 4 d x = .
3
1

= E( X 2)

[ E( X) ]2 =

E ( X 10 ) does NOT exist since

10

5 5

3 4

5
.
48

5 x 6 d x diverges.

Median:

FX( m ) =

30th percentile:

1
.
2

F X ( 0.30 ) = 0.30.

1 m5 =

1
.
2

m=

2 1.1487.

1 ( 0.30 ) 5 = 0.30.

0.30 =

1
1.07394.
0.70

Example 4:

( Standard ) Cauchy distribution:

fX( x ) =

1+ x 2

),

< x < .

Even though f X ( x ) is symmetric about zero, E ( X ) is undefined since

1
dx = .
x
2

1
+
x

FX( x ) =

2
1 + y

dy =

arctan( x ) + ,
1
2

< x < .

P ( X < 1 ) = P ( 1 < X < 0 ) = P ( 0 < X < 1 ) = P ( X > 1 ) = 0.25.

M X ( t ) is undefined for all t 0.

M X ( 0 ) = 1.

Theorem 1:

M X ( t ) = M X ( t ) for some interval containing 0

1

Theorem 2:

M X' ( 0 ) = E ( X )

fX1( x ) = fX2( x )

M X" ( 0 ) = E ( X 2 )

(k)
MX ( 0) = E( X k)

Theorem 3:

Let Y = a X + b. Then M Y ( t ) = e b t M X ( a t )

Example 5:
Suppose a discrete random variable X has the following probability distribution:
P( X = 0 ) = p,
a)

P( X = k ) =

1
, k = 1, 2, 3,
k
2 k!

Find the value of p that would make this a valid probability distribution.

1
= 1.
k
k =1 2 k !

p+

Must have

Since

1
ak
= e 1 2 1.
= ea,

k
k =0 k !
k =1 2 k !

Therefore, p + ( e 1 2 1 ) = 1

b)

p = 2 e1 2.

and

Find E ( X ).

E(X) =

x p ( x)

= 0 2e

)+
k =1

12

all x

c)

2 k =1 2 k 1 ( k 1 )!

2 n=0

1
=
k
2 k k!

1
=
2 n n!

e1 2

k
k =1 2 ( k 1 )!
.

Find the variance of X, Var ( X ).

E(X(X 1)) =

k =2
=

1
=
k ( k 1 )
2 k k!
1

4 k = 2 2 k 2 ( k 2 )!

4 n=0

2 k ( k 2 )!
1

k =2

1
=
2 n n!

e1 2
4

E( X2) = E( X (X 1) ) + E( X) =

e1 2 .

Var ( X ) = E ( X 2 )

3
[E ( X )]2 = e1 2

e.

d)

Find the moment-generating function of X, M X ( t ).

MX(t ) =

e t x p ( x) = 1 2 e 1 2

all x

) + e t k k1
2 k!
k =1

t
e

(2 e1 2 ) +

k!

k =1

(2 e1 2 ) +

et 2

e
1

= 1 e1 2 + e e 2 .

e)

Use the moment-generating function of X, M X ( t ), to find E ( X ).

M X' ( t ) =

f)

ee

t
e 2 ,

E ( X ) = M X' (0 ) = e

12

Use the moment-generating function of X, M X ( t ), to find the variance of X, Var ( X ).

t
M '' ( t ) = e e
X

t
e 2

E ( X 2 ) = M X'' (0 ) =
Var ( X ) = E ( X 2 )

ee

t
e 2 ,

e1 2.

3
[E ( X )]2 = e1 2
4

1
4

e.

Example 6:
Let a > 2. Suppose a discrete random variable X has the following probability
distribution:

p ( 0 ) = P ( X = 0 ) = c,
1
,
k
a

p(k) = P( X = k ) =
a)

k = 1, 2, 3, .

Find the value of c ( c will depend on a ) that makes this is a valid probability
distribution.

Must have

p(x ) = 1.

1
= 1.
k =1 a k

c+

all x

bk

k =0

1
1 b

| b | < 1.

1
1
1
1
.
=
k k 1 = 1 1 =

1
1
a
a
k =1
k = 0

a
OR

1
1 1
1
1
1
.
=
=
=
k =1 a k a k = 0 a k a 1 1 a a 1

c+

b)

a 1

= 1.

c = 1

a
a2
.
= 2
a 1
a 1 a 1
1

Find P ( odd ).

P ( odd ) = p ( 1 ) + p ( 3 ) + p ( 5 ) + =

1
=

c)

first term
=
1 base

a
1

a
a 2 1

a2

Find the moment-generating function of X, M X ( t ). For which values of t does

it exist?

MX(t) = E(et X) = 1 c +

bk

k =0

1
1 b

k =1

1
.
ak

| b | < 1.

etk k =
a
k =1
k = 0

etk

t
e
a

k
1
et
a

1
1
1.
=

t
t aet

e
e
a

1
a

OR
t
tk 1 = e
e

a k a k =0
k =1

Need

et
a

MX(t) =

< 1.

et

k
=

et

a 1

1
et

et
aet

t < ln a.

et
a2
a
1
,
+
=

a 1 a e t a e t a 1

t < ln a.

a
aet

1.

d)

Find E ( X ).

E ( X ) = M 'X ( 0 ).

et
d a 2
M 'X ( t ) =
+
dt a 1 a e t

e t a e t e t e t

a e t

a e t
a e t

OR
M 'X ( t ) =

1
d a
a e t
a
e t =
.
=

2
2

dt a e t a 1
t

a e t
ae

E ( X ) = M 'X ( 0 ) =

( a 1 ) 2

OR
k 1

a 1
1
1
1
1 a

E(X) = k

E(Y ) ,
k
=
=
a 1
a k a a 1 k = 1 a
k =1
a

where Y is a Geometric random variable with probability of success

a 1
a

E(Y ) =

a
a 1

Therefore, E ( X ) =

( a 1 ) 2

OR

E ( X ) = 1

a
1

E(X) =

a
1

+ 3

a
1

+ 2

a
+ 3

+ 4

+ 5

1
5

a
+ 4

+ 6

1
6

a
+ 5

+ ...

+ ...

1
1
1
1
1
1
1
1
1
.
+
+
+
+
+
+ ... =
=
1 E ( X ) =
k
a

1
a
a1 a 2 a 3 a 4 a 5 a 6
k =1 a

Therefore, E ( X ) =

e)

+ 2

( a 1 ) 2

Find the cumulative distribution function F ( x ) = P ( X x ).

If x < 0,

F ( x ) = P ( X x ) = 0.

If k = 0, 1, 2, 3, ,

1
1
1
1
=
=

n
a k ( a 1 )
a k +1 1 1 a
n = k +1 a

1 F(k) = P(X > k) =

F(k) = 1

a k ( a 1 )

Since X is a discrete integer-valued random variable, if k x < k + 1,

F(x) = F(k) = 1

a k ( a 1 )

Therefore,

F(x) =

1
1 k
a ( a 1)

x<0
k x < k +1

k = 0, 1, 2, 3, ...

Example 7:
Let > 0. Suppose the probability density function of X is f X ( x ) = e

, x > 0.

( Exponential distribution. )
a)

Find the moment-generating function of X.

MX( t ) = E( e t X ) =

e t x f (x ) dx =

t x e x dx

x( t )

e
(
)
x
t
= e
=
dx =

b)

t < .

Use the moment-generating function of X to find E ( X ).

'

MX(t ) =

( t )2

t < .

1
E( X ) = MX(0 ) =
.

'

Example 8:
Let X be a discrete Binomial ( n, p ) random variable. That is, suppose the p.m.f.
of X is

n
p X ( k ) = p k ( 1 p ) n k ,
k

MX(t ) =

k =0

k = 0, 1, 2, , n.

e t k p k ( 1 p ) n k

k
n
p e t ( 1 p ) n k =
k

k =0

[(1 p) + p et]

Example 9:
Let X be a discrete Geometric ( p ) random variable. That is, suppose the
x 1 p, x = 1, 2, 3, .
probability mass function of X is p X ( x ) = ( 1 p )
a)

Find the moment-generating function of X.

MX(t ) =

e t k ( 1 p ) k 1 p = p e t

e t (k 1) (1 p ) k 1
k =1

k =1

= p e t

( 1 p ) e t

n=0

b)

n
=

p e t
,
1 (1 p ) e t

t < ln ( 1 p ).

Use the moment-generating function of X to find E ( X ).

p e t 1 (1 p ) e t

'

MX(t ) =

2
1 (1 p ) e t

'

t
t
p e (1 p ) e

(1 (1 p ) e t ) 2

p e t

E( X ) = MX(0 ) =

t < ln ( 1 p ).

p
1
=
.
( p )2
p

Example 10:
Let X be a random variable distributed uniformly over the interval [ a , b ].

MX( t ) = E( e t X ) =

b
t x f ( x ) dx =

et x b

b a t a
1

M X ( 0 ) = 1.

e
a

etb et a ,
t (b a )

tx

ba

t 0.

dx

Example 11:
Let X be a Poisson ( ) random variable. That is,
P(X = k) =
a)

k e
,
k!

k = 0, 1, 2, 3, .

Find the moment-generating function of X, M X ( t ).

k e
t
k
MX(t ) =
=
e
k
!
k =0

e e e

1.9.17 in the 7th edition

e (e

e t

k =0
1)

k!

( 1.9.16 in the 6th edition )

Let ( t ) = ln M ( t ), where M ( t ) is the m.g.f. of a distribution.

Prove that X' ( 0 ) = and X" ( 0 ) = . The function ( t ) is
2

called the cumulant generating function.

( ln M X ( t ) ) '
Since

M X' ( t )

( ln M X ( t ) ) "

MX (t )

M X ( 0 ) = 1,

M X ' ( 0 ) = E ( X ),

M X" ( t ) M X ( t ) M X' ( t )

[ MX (t ) ]

M X " ( 0 ) = E ( X 2 ),

X' ( 0 ) = ( ln M X ( t ) ) ' | t = 0 = E ( X ) = X
X" ( 0 ) = ( ln M X ( t ) ) " | t = 0 = E ( X 2 ) [ E ( X ) ] 2 = Var ( X ) = X2
b)

Find E ( X ) and Var ( X ).

ln M X ( t ) = ( e t 1 ).

( ln M X ( t ) ) ' = e t.

( ln M X ( t ) ) ' | t = 0 = E ( X ) = .

( ln M X ( t ) ) " = e t.

( ln M X ( t ) ) " | t = 0 = Var ( X ) = .

Example 12:
Let Y denote a random variable with probability density function given by
1
2

f(y) =
a)

< y < .

( double exponential p.d.f. )

Find the moment-generating function of Y. For which values of t does

it exist?

0
ty 1 y
MY(t ) = e e
dy =
2

1 0
=
2

ty 1 y
ty 1 y
dy
e e
dy + e e
2
2
0

ty 1 y
ty 1 y
dy
e dy + e e
2
2
0

y (t +1 )

1 y ( t 1 )
dy
dy + e
2 0

Note that the first integral converges only if t + 1 > 0,

and the second integral converges only if t 1 < 0.
Therefore, the moment-generating function is only defined for 1 < t < 1.
MY(t ) =
=

b)

y(t +1 ) 0
y ( t 1 )
1
1
1
1
+
=
e
e

0
2 ( t +1 )
2 ( t 1 )
2 ( t +1 ) 2 ( t 1 )

( t 1 ) ( t +1 )
2 ( t + 1 )( t 1 )

2 t 2 1

1
1 t 2

1 < t < 1.

Find E( Y ).
M Y' ( t ) = ( 1 t 2 ) 2 ( 2 t ) = 2 t ( 1 t 2 ) 2

E ( Y ) = M Y' ( 0 ) = 0.
OR

E( Y ) =

1
y 2 e

dy = 0,

since y

1 y
e
2

is an odd function.

c)

Find Var( Y ).

M Y" ( t )

2 2

= 2(1 t )

2 3

+ 2 t(2)(1 t )

E ( Y 2 ) = M Y" ( 0 ) = 2.

(2t) =

2 + 6t 2

(1 t 2 ) 3

Var( Y ) = 2 0 2 = 2.

OR
2

Var( Y ) = E ( Y ) [ E ( Y ) ] =

d)

2 1
y 2e

dy = = 2.

Find the cumulative distribution function F ( y ) = P ( Y y ).

If y < 0,

F(y) =

If y 0,

F(y) =

1 x
e
dx =
2

1 x
dx =
e
2

1 x
1 y
e dx = e .
2
2

1 x
e dx +
2

y
1 1
1 y
= 1 e
.
+ 1 e
2 2
2

Therefore,

1 y
e

F(y) =

1 y
1 e
2

y<0
y0

1 x
e dx
2

Example 13:
A simple model for describing mortality
in the general population in a particular
country is given by the probability density
function

f (y) =

252
10

18

y 6 ( 100 y ) 2 ,
0 < y < 100.

a)

Verify that f ( y ) is a valid probability

density function.
1.

f ( y ) 0 for each y;

2.

f ( y ) dy = 1 .

100

y ( 100 y ) dy = 252 x ( 1 x ) 2 dx

f ( y ) dy = 10 18
252

1 8 1 9 1
2
1 7
= 252 x 2 x + x = 252
= 1.
8
9
504
7
0

b)

Based on this model, which event is more likely

or

A: a person dies between the ages of 70 and 80

B: a person lives past age 80?

80

A:

252

10 18

70

y ( 100 y ) dy =
6

0.8

252 x

( 1 x ) 2 dx

0.7

1 8 1 9 0.8
1 7
= 252 x 2 x + x
8
9
7
0.7
0.7382 0.4628 = 0.2754.

100

B:

y ( 100 y ) dy =

252

10 18

1.0

252 x

( 1 x ) 2 dx

0.8

80

1 8 1 9 1 .0
1 7
= 252 x 2 x + x
8
9
7
0.8
1 0.7382 = 0.2618.

A is more likely.

c)

Given that a randomly selected individual just celebrated his 60th birthday, find
the probability that he will live past age 80.
100

P ( over 80 | over 60 ) =

252

10
P ( over 80 over 60 )
= 80
100
P ( over 60 )
252

60 10

d)

18

18

y 6 ( 100 y ) 2 dy
y 6 ( 100 y ) 2 dy

1 0.7382
0.2618
=
0.3408.
1 0.2318
0.7682

Find the value of y that maximizes f ( y ) ( mode ).

f '(y) =

5
6
252
6 y ( 100 y ) 2 2 y ( 100 y )

18
10

252
10

252
10

18

18

y ( 100 y ) [ 6 ( 100 y ) 2 y ]
5

y ( 100 y ) [ 600 8 y ] = 0.
5

y = 0, y = 100 (not max), y = 75 years (max).

e)

Find the (average) life expectancy.

100

y f ( y ) dy =

E( Y ) =

252

0 10

18

y ( 100 y ) dy = 252 100 x 7 ( 1 x ) 2 dx

2

1
1
2
1
1
= 252 100 x 8 2 x 9 + x 10 = 252 100
= 70 years.
9
10
720
8
0
OR
Y
.
100

Consider X =

Then Y = 100 X, and X has the probability density function

f ( x ) = 252 x 6 ( 1 x ) 2 , 0 < x < 1.

Then X has Beta distribution with = 7 and = 3.
E( X ) =

f)

7
= 0.70.
7+3

E ( Y ) = 100 E ( X ) = 70 years.

Find the standard deviation of the lifetimes.

Var ( X ) =

73
11 10 2

21
.
1100

Var ( Y ) = 100 2 Var ( X ) =

SD ( Y ) 13.817.
OR

E( Y ) =

f ( y ) dy =

100

252

0 10

18

y 8 ( 100 y ) 2 dy =

Var ( Y ) = E ( Y 2 ) [ E ( Y ) ] 2 =
SD ( Y ) =

Var (Y ) =

2100
.
11