Pairs of Random Variables

Pairs of Random Variables

In this chapter :
5-1 Two Random Variables
5-2 Pairs of Discrete Random Variables
5-3 Marginal PMF
5-4 Joint cdf of X and Y
5-5 Joint pdf of two continuous RV
5-6 Independence of 2 Rand. Var.
5-7 Expected Value of a function of 2 RV
5-8 Joint Moments
5-9 Covariance and Correlation Coefficient
 Two Random Variables
• Two quantities are of interest.
• Function assigning a pair of real numbers to
each random experiment’s outcome.
• For each  in S, X() = (X(), Y()) is defined.
• Example:  is a student. H() and W() are
respectively height and weight of the student
chosen randomly.
 Examples of RV specifications
• A = {X + Y ≤ 10}
• B = {min (X, Y) ≤ 5}
• C = {X2 + Y2 ≤ 100}
 Scattergram
• Scattergram:
• dot at every observation pair (x, y).
• used to deduce the joint behavior of two
random variables (X, Y).
Pairs of Discrete Random Variables
• X = (X, Y) take values for a countable set
SX,Y = {(xj, yk), j = 1, 2, …, k = 1, 2, …}
• The joint pmf of X specifies probabilities of
{X = x} and {Y = y}: pX,Y = P[{X = x}  {Y = y}]
= P[X = x, Y = y]
 Marginal PMF
• pX(xj) = P[X = xj]
= P[X = xj and Y = anything]
= P[

{X = xj and Y = y1}  {X = xj and Y = y2}  …]

= k 1 p X ,Y  x i , y k 
Example
 Figure (5-6)
• 2 “loaded” dice
y
1/42 1/42 1/42 1/42 1/42 2/42
6
1/42 1/42 1/42 1/42 2/42 1/42
5
1/42 1/42 1/42 2/42 1/42 1/42
4
1/42 1/42 2/42 1/42 1/42 1/42
3
2 1/42 2/42 1/42 1/42 1/42 1/42

1 2/42 1/42 1/42 1/42 1/42 1/42 x

1 2 3 4 5 6
 Joint cdf of X and Y
• Definition: FXY(x1, y1) = P[X ≤ x1, Y ≤ y1]
• Properties:
– If x1 ≤ x2 and y1 ≤ y2 then FXY(x1, y1) ≤ FXY(x2, y2)
– FXY(x1, -) = 0, FXY(-, y1) = 0
– FXY(, ) = 1
– Marginal cdf: FX(x1)= FXY(x1, ) and FY(y1)= FXY(, y1)
– P[x1 < X ≤ x2, y1 < Y ≤ y2] = FXY(x2, y2) – FXY(x2, y1)
- FXY(x1, y2) + FXY(x1, y1)
 Proof of last property
P[B] = P[x1< X ≤x2, Y ≤ y1] = FXY(x2, y1) - FXY(x1, y1)
Note that FXY(x2, y2) = P[A] + P[B] + FXY(x1, y2)
Solve for P[B]
y x1 x2 x y x1 x2 x
y2
y1 y1 A
B B
 Example
 1  e  x   1  e   y  x  0, y  0
FX Y x,y  
0 elsewhere

Find marginal cdf’s.

FX  x   lim FX Y  x , y   1  e  x x 0
y 

FY  y   lim FX Y  x , y   1  e   y y 0
x 
 Example (continued)
Find probability of A = {X ≤ 1, Y ≤ 1},
B = {X > x, Y > y} (x and y positive),
D = {1 < X ≤ 2, 2 < Y ≤ 5}
Solution:
P[A] and P[D] immediate.
Bc = {X ≤ x}  {Y ≤ y} 
P[Bc] = P[X ≤ x] + P[Y ≤ y] - P[X ≤ x, Y ≤ y]
P[Bc] = (1 – e-x) + (1 – e-y) - (1 – e-x) (1 – e-y)
P[B] = 1 - P[Bc] = e-x e-y
 Joint pdf of two continuous RV
• fXY(x, y) non-negative function
• For each event B
P [ X in B ]   f (x , y )dxdy XY
B

• Entire plane  
1    f (x , y )dxdy XY
 

• cdf and pdf

x y
 2 FX Y (x , y )
FX Y (x , y )   f
 
XY (x ', y ')dx 'dy ' f XY (x , y ) 
x y
 Example: Jointly uniform RV
1 0  x  1, 0  y  1
f XY  x , y   
0 elsewhere

Find the joint cdf.

Discuss five regions.
x < 0 and y < 0: FXY(x, y) = 0 x y
FX Y  x , y     1dx ' dy '  xy
x and y in the unit interval: 0 0
x 1
x in the unit interval, y > 1: FX Y  x , y     1dx ' dy '  x
y in the unit interval, x > 1: 0 0
1 y
FX Y  x , y     1dx ' dy '  y
x > 1, y > 1: FXY(x, y) = 1 0 0
 Independence of 2 Rand. Var.
• Definition:
P[X in A1, Y in A2] = P[X in A1]P[Y in A2]
• In general
FXY(x, y) = FX(x)FY(y) for all x and y
fXY(x, y) = fX(x)fY(y) for all x and y
 Example
ce  x e  y 0  y  x <
f XY x,y  
0 elsewhere

Find c.
x 
c
1    ce e dxdy   ce
x y x
 1e x
 dx  2
0 0 0

c = 2.
 Example, continued
• Find marginal 
pdf’s. x
f X  x    f X Y  x , y  dy   2e  x e  y dy  2e  x  1  e  x 
0 0

for x  [0, [.

 
f Y  y    f X Y  x , y  dx   2e  x e  y dx  2e 2 y
0 y

for y  [0, [.

Note that X and Y are NOT independent.
 Expected Value of a function of 2 RV
Z  g  X ,Y 
 
EZ     g  x , y  f  x , y  dxdy
XY
 

• Verify that E(X + Y) = E(X) + E(Y).

 Joint Moments
• The jkth joint moment 
of X and Y:
E  X jY k     x j y k f X Y  x , y  dxdy
 
• If j = k = 1, E[XY] is the CORRELATION of X and
Y. If E[XY] = 0, X and Y are orthogonal.
• The jkth central moment of X and Y:
E[(X – E(X))j(Y – E(Y))k]
• Covariance of X and Y:
COV(X, Y) = E[(X – E(X)) (Y – E(Y))]
 Covariance
• Prove that
COV(X, Y) = E[XY] – E[X]E[Y].
• Prove that COV of independent random
variables is zero.
• Solution:
COV(X, Y) = E[(X – E(X)) (Y – E(Y))]
COV(X, Y) = E[X – E(X)]E[Y – E(Y)] = 0
 Correlation Coefficient
COV  X ,Y 
X Y 
 X Y

• Note that -1 ≤ XY ≤ 1.

• Interpretation.
 Example
2e  x e  y 0  y  x <
f XY x,y  
0 elsewhere

• Find E[XY], COV(X,Y) and XY.

• Reminder:
f X  x   2e  x  1  e  x  f Y  y   2e 2 y

• Verify that E[X] = 3/2 and VAR(X) = 5/4.

• Verify that E[Y] = 1/2 and VAR(Y) = 1/4.
 Example
x 
E  X Y     xy 2e  x e  y dxdy   2xe  x  1  e  x  xe  x  dx  1
0 0 0

COV(X, Y) = E[XY] – E[X]E[Y]

= 1 – 3/2*1/2 = 1/4
XY = 1/(5)1/2