UNIT 2 - Rejinpaul PDF

www.rejinpaul.
com
www.rejinpaul.com
18 December 2014 Two Dimensional Random 1

Variables by Dr M Radhakrishnan
Download Useful Materials from Rejinpaul.com

www.rejinpaul.com
www.rejinpaul.com
MA 6453 PROBABILITY AND QUEUEING THEORY

(Regulation 2013)
UNIT SYLLABUS
UNIT II TWO - DIMENSIONAL RANDOM VARIABLES
Joint distributions – Marginal and conditional

distributions – Covariance – Correlation and Linear
regression – Transformation of random variables


www.rejinpaul.com
www.rejinpaul.com
JOINT DISTRIBUTIONS
&
MARGINAL AND CONDITIONAL DISTRIBUTIONS


www.rejinpaul.com
www.rejinpaul.com
TWO – DIMENSIONAL RANDOM VARIABLES

Let S be the sample space associated with a random experiment E.
Let X = X(s) and Y = Y(s) be two functions each assigning a real
number to each outcomes s ∈ S. Then (X, Y ) is called a
two - dimensional random variable.
Examples :
1. Signal transmission : X is high quality signals and
Y low quality signals.


www.rejinpaul.com
www.rejinpaul.com
TWO – DIMENSIONAL DISCRETE RANDOM VARIABLES

If the possible values of (X, Y) are finite or countably infinite,
then (X, Y) is called a two - dimensional discrete random variables.
Example:
(Sample space for dice) Consider the experiment of
tossing two fair dice. The sample space for this experiment
has 36 equally likely points. Let X = sum of the two dice
and Y = |deference of two dice|. The random variable (X, Y )
defined above is called a discrete random vector because it
has only a countable (in this case, finite) number of
possible values.


www.rejinpaul.com
www.rejinpaul.com
JOINT PROBABILITY MASS FUNCTION (PMF)
When both X and Y are discrete random variables,

we define their joint PMF or simply the Probability
Function of (X, Y) as follows :
p XY ( x i , y j ) = P [ X = x i ,Y = y j ]
provided the following conditions are satisfied.
(i ) 0 ≤ p XY ( x i , y j )) ≤ 1 ∀ i , j
(ii ) ∑∑p
i j
XY (xi , y j ) = 1


www.rejinpaul.com
www.rejinpaul.com
MARGINAL PROBABILITY DISTRIBUTION

(For discrete random variables)
If (X, Y) is a two - dimensional discrete RV, then the

probability distribution of X, also called marginal
probability mass function of X, denoted by p X (x i ), is defined as
p X (x i ) = ∑ p XY (x i , y j ) = P[X = x i ]
j
Similarly, the marginal probability mass function of Y, denoted by

p Y (y j ), is defined as
p Y (y j ) = ∑ p XY (xi , y j ) = P[Y = y j ]
j


www.rejinpaul.com
www.rejinpaul.com
CONDITIONAL PROBABILITY DISTRIBUTION

Let (X, Y) be a discrete two - dimensiona l discrete random
variables with the joint PMF p XY ( x , y ). Then the conditional
PMF of Y, given X = x , is given by
P[ X = x ,Y = y ] p XY ( x , y )
pY X ( y x ) = = provided p X ( x ) > 0.
P[ X = x ] pX ( x )
Similarly,
the conditional PMF of X, given Y = y , is given by
P[ X = x ,Y = y ] p XY ( x , y )
PX Y ( x y ) = = provided pY ( y ) > 0.
P[Y = y ] pY ( y )


www.rejinpaul.com
www.rejinpaul.com
INDEPENDENT RANDOM VARIABLES

Let (X, Y) be a two - dimensiona l discrete random variable.

Then X and Y are independent random variables iff
PX ,Y ( x , y ) = PX ( x ). PY ( y )
where PX ,Y ( x , y ) is the joint p.m.f of (X, Y) and PX ( x ) and
PY ( y ) are the marginal p.m.f of X and Y.


www.rejinpaul.com
www.rejinpaul.com
CONDITIONAL MEANS
If X and Y are discrete random variables with the

joint PMF p XY ( x , y ), the conditiona l expected value of Y,
given that X = x , is defined by E [Y X = x ] = ∑ y pY X (y x)
y
Similarly,
the conditional expected value X, given Y = y ,
is given by E [ X Y = y ] = ∑ x p X Y ( x y )
x


www.rejinpaul.com
www.rejinpaul.com
Example :
The joint probability mass function of (X , Y ) is given by
p(x , y ) = k (2 x + 3y ), x = 0,1,2; y = 1,2,3. Find all the
marginal and conditional probability distributi ons.
Solution :
The joint probability mass function of ( X , Y ) is given below
Y 1 2 3
X
0 3k 6k 9k
1 5k 8k 11k
2 7k 10k 13k


www.rejinpaul.com
www.rejinpaul.com
Since p( x , y ) is a probability mass function, we have

∑ p( x , y ) = 1
3 k + 6 k + 9 k + 5 k + 8 k + 11 k + 7 k + 10 k + 13 k = 1
1
72 k = 1 ⇒ k = .
72
Marginal probability distribution of X
P [ X = 0] = 3k + 6k + 9k = 18k =
18 1
=
72 4
P [ X = 1] = 5k + 8k + 11k = 24k =
24 1
=
72 3
P [ X = 0] = 7 k + 10k + 13k = 30k =

30 5
=
72 12


www.rejinpaul.com
www.rejinpaul.com
Marginal probabilit y distribution ofY
P [Y = 1] = 3k + 5k + 7k = 15k =
15 5
=
72 24
[ ]
P Y = 2 = 6k + 8k + 10k = 24k =
24 1
=
72 3
[ ]
P Y = 3 = 9k + 11k + 13k = 33k =
33 11
=
72 24


www.rejinpaul.com
www.rejinpaul.com
Conditiona l distributi on of X given Y = 1
P [ X = 0 , Y = 1] 3k
[ ]
P X =0 Y = 1 = = =
3 1
=
P [Y = 1] 15k 15 5
P [ X = 1 , Y = 1] 5k
P [ X = 1 Y = 1] =
5 1
= = =
P [Y = 1] 15k 15 3
P [ X = 2 , Y = 1] 7k
P [ X = 2 Y = 1] =
7
= =
P [Y = 1] 15k 15


www.rejinpaul.com
www.rejinpaul.com
Conditiona l distributi on of X given Y = 2

P [ X = 0 , Y = 2]
P [ X = 0 Y = 2] =
6k 6 1
= = =
P [Y = 2] 24 k 24 4
P [ X = 1 , Y = 2]
P [ X = 1 Y = 2] =
8k 8 1
= = =
P [Y = 2] 24k 24 3
P [ X = 2 , Y = 2] 10 k
P [ X = 2 Y = 2] =
5
= =
P [Y = 2] 24 k 12


www.rejinpaul.com
www.rejinpaul.com
Conditional distribution of X given Y = 3

P [ X = 0 , Y = 3]
P [ X = 0 Y = 3] =
9k 9 3
= = =
P [Y = 3] 33k 33 11
P [ X = 1 , Y = 3] 11k 11 1
P [ X = 1 Y = 3] = = = =
P [Y = 3] 33k 33 3
P [ X = 2 , Y = 3] 13k 13
P [ X = 2 Y = 3] = = =
P [Y = 3] 33k 33


www.rejinpaul.com
www.rejinpaul.com
Conditiona l distributi on of Y given X = 0

P [ X = 0 , Y = 1] 3k
P [Y = 1 X = 0] =
3 1
= = =
P [ X = 0] 18k 18 6
P [ X = 0 , Y = 2] 6 k
[
P Y =2 X = 0 =] = =
6 1
=
P [ X = 0] 18k 18 3
P [ X = 0 , Y = 3] 9 k
P [Y = 3 X = 0] =
9 1
= = =
P [ X = 0] 18k 18 2


www.rejinpaul.com
www.rejinpaul.com
Conditiona l distributi on of Y given X = 1

P [ X = 1 , Y = 1] 5k
P [Y = 1 X = 1] =
5
= =
P [ X = 1] 24k 24
P [ X = 1 , Y = 2]
P [Y = 2 X = 1] =
8k 8 1
= = =
P [ X = 1] 24k 24 3
P [ X = 1 , Y = 3]
P [Y = 3 X = 1] =
11k 11
= =
P [ X = 1] 24k 24


www.rejinpaul.com
www.rejinpaul.com
Conditional distribution of Y given X = 2

P [ X = 2 , Y = 1] 7k
[
P Y =1 X = 2 =] = =
7
P [ X = 2] 30k 30
P [ X = 2 , Y = 2] 10k 10 1
P [Y = 2 X = 2] = = = =
P [ X = 2] 30k 30 3
P [ X = 2 , Y = 3] 13k 13
P [Y = 3 X = 2] = = =
P [ X = 2] 30k 30


www.rejinpaul.com
www.rejinpaul.com
Example :
The Joint p.m.f of X and Y is
Y
p(x,y)
0 1 2
0 0.1 0.04 0.02
X 1 0.08 0.20 0.06
2 0.06 0.14 0.30
Determine if X and Y are independent.


www.rejinpaul.com
www.rejinpaul.com
Solution :
The marginal p.m.f of X and Y are given by
Y
p(x,y) p(x) = P(X=x)
0 1 2
0 0.1 0.04 0.02 0.16
X 1 0.08 0.20 0.06 0.34
2 0.06 0.14 0.30 0.5
p(y) = P(Y=y) 0.24 0.38 0.38 1
If p XY(x,y) = p X(x).p Y(y) for all x and y,

we can say that X and Y are independent.


www.rejinpaul.com
www.rejinpaul.com
We have p(X = 0 )× p(Y = 0 ) = 0.16× 0.24 = 0.0384
≠ p(x = 0 ,y = 0 ) = 0 .1
Similarly we can verify that

p(X = 1 )× p(Y = 1 ) ≠ p(x = 1,y = 1 )
p(X = 2 )× p(Y = 2 ) ≠ p(x = 2 ,y = 2 ) and so on.
Hence the RVs X and are not independent.


www.rejinpaul.com
www.rejinpaul.com
TWO – DIMENSIONAL CONTINUOUS RANDOM VARIABLES

If (X, Y) can assume all values in a specified region R
in the xy - plane, (X, Y) is called a two - dimensional
continuous random variables.
Example:
Suppose X denotes the duration of an eruption (in second)
of Volcano, and Y denotes the time (in minutes) until the
next eruption. We might want to know if there is a
relationship between X and Y. Or, we might want to know
the probability that X falls between two particular
values a and b, and Y falls between two particular
values c and d. That is, we might want to
know P(a < X < b, c < Y < d ). In this case (X, Y) is called two
dimensional continuous random variables.

www.rejinpaul.com
www.rejinpaul.com
JOINT PROBABILITY DENSITY FUNCTION (PDF)
If (X, Y) is a two - dimensional continuousrandom variable such that

 dx dx dy dy 
Px − ≤ X ≤ x + , y − ≤ Y ≤ y +  = f ( x, y )dxdy,
 2 2 2 2 
then f ( x, y ) is called the joint p.d.f of (X, Y), provided f ( x, y )
satisfies the following conditions:
( i ) f ( x, y ) ≥ 0 for all ( x, y ) ∈ R, whre R is the range space.
(ii) ∫∫ f(x, y)dxdy = 1
R
( iii ) If D is a subspace of the range space R, then

P{(X, Y) ∈ D} = ∫∫ f(x, y)dxdy.
D


www.rejinpaul.com
www.rejinpaul.com
MARGINAL PROBABILITY DISTRIBUTION

(For continuous random variables)
Let (X, Y) be the two - dimensional continuous RV.
Then the marginal density function of X is denoted
by f X ( x ) and is defined as,
∞
f X ( x) = ∫f
−∞
XY ( x , y )dy
Similarly,
the marginal density function of Y is denoted by
fY ( y ) and is defined as,
∞
fY ( y ) = ∫f
−∞
XY ( x, y )dx


www.rejinpaul.com
www.rejinpaul.com
CONDITIONAL PROBABILITY DISTRIBUTION

Consider two continuous random variables X and Y

with the joint PDF f XY ( x , y ).
The conditiona l PDF of Y, given X = x , is given by
f ( x, y)
f Y X ( y x ) = XY provided f X ( x ) > 0
f X ( x)
Similarly,
the conditional PDF of X, Y = y , is given by
f ( x, y)
f X Y ( x y ) = XY provided f Y ( y ) > 0
fY ( y )


www.rejinpaul.com
www.rejinpaul.com
INDEPENDENT RANDOM VARIABLES

Two random variables X and Y with joint p.d.f f(x, y)

and marginal p.d.fs f(x) and f(y) respective ly are said
to be independen t iff
f X Y ( x y ) = f X ( x ) f Y ( y ).


www.rejinpaul.com
www.rejinpaul.com
Example :
cx ( x − y ) ; 0 < x < 2 , − x < y < x
Given f XY ( x , y ) =  .
0 ; otherwise
(i ) Evaluate c (ii ) f X ( x ) (iii ) fY X ( y x ) and (iv ) fY (y ).
Solution :
Given f ( x , y ) is the jo int p.d . f , we have
∫∫ f ( x , y ) dx dy = 1
2 x
c∫ ∫ (x − xy ) dy dx = 1
2
0 −x
c ∫  x 2 ( x − (− x )) − ( x 2 − x 2 ) dx = 1
 x 
0  
2


www.rejinpaul.com
www.rejinpaul.com
c
[16 − 0]= 1 ⇒ 8c = 1 ⇒ c = 1
2 8
∴ f ( x , y ) = ( x − xy ) ; 0 < x < 2 , − x < y < x

1 2
8
∞ x
f X (x ) = ∫ f (x , y ) dy = ∫ (x 2 − xy )dy
1
−∞
8 −x
1  2 x 
y 
=  x 2 (y )− x − x   
x
8   2  − x 


www.rejinpaul.com
www.rejinpaul.com
=  x 2 (x − (− x )) − (x 2 − x 2 )
1 x 
8 2 
= [x (2x ) − 0]
1 2
8
x3
= ,0 < x < 2
4
(
1 2
f (x , y ) 8 x − xy
)
f Y X (y x ) = =
f X (x ) x3
4
4 x (x − y )
=
8 x3
x−y
= 2
,− x< y < x
2x


www.rejinpaul.com
www.rejinpaul.com
Example :
8 x y ; 0 < x < y < 1
Given the joint pdf of (X, Y) as f ( x , y ) =  . Find the
0 , otherwise
marginal and conditional probability density functions of X and Y.
Are X and Y are independent?
Solution :
Marginal density of X is
1
f X ( x ) = ∫ f ( x , y ) dy = ∫ 8 x y dy
x
1
 y2 
1
= 8 x ∫ y dy = 8 x 
x  2 x
= 4 x (1 − x 2 ) , 0 < x < 1


www.rejinpaul.com
www.rejinpaul.com
Marginal density of Y is
y
fY ( y ) = ∫ f ( x , y ) dx = ∫ 8 x y dx
0
y y
x  2
= 8 y ∫ x dx = 8 y  
0  2 0
= 4 y ( y 2 − 0) = 4 y 3 , 0 < y < 1
and f X ( x ) . fY ( y ) = 4 x (1 − x 2 ). 4 y 3 ≠ 8 xy = f XY ( x , y )
Therefore X and Y are not independent.
Conditional density of X given Y is

f (x , y) 8 x y 2 x
f X Y (x y) = = = 2 , 0< x < y
fY ( y ) 4y 3
y


www.rejinpaul.com
www.rejinpaul.com
Conditional density of Y given X is

f (x , y )
f Y X (y x ) =
f X (x )
8xy
=
4 x (1 − x 2 )
2y
= , x< y <1
1− x 2


www.rejinpaul.com
www.rejinpaul.com
JOINT CUMMULATIVE DISTRIBUTION FUNCTION (CDF)
If (X, Y) is a two - dimensional RV(discrete or continuous), then by

F ( x , y ) = P[ X ≤ x , Y ≤ y ] is called the cdf of (X, Y).
In the discrete case,
F(x, y) = ∑ ∑ p( x , y )
j i
i j
yj≤ y xi ≤ x
In the continuous case,

y x
F(x, y) = ∫ ∫ f(x, y)dxdy

-∞ -∞


www.rejinpaul.com
www.rejinpaul.com
PROPERTIES OF JOINT CDFs
1. 0 ≤ F XY ( x , y ) ≤ 1 for − ∞ < x < ∞ , − ∞ < y < ∞ .
2. lim FXY ( x , y ) = FXY (∞, ∞ ) = 1

x →∞
y →∞
3. lim FXY ( x , y ) = FXY ( −∞ , y ) = 0

x → −∞
4. lim FXY ( x , y ) = FXY ( x ,−∞ ) = 0

y → −∞


www.rejinpaul.com
www.rejinpaul.com
MARGINAL CDFs
The marginal distribution of X with respect to the joint CDF
FXY ( x , y ) is FX ( x ) = P[ X ≤ x] = P[ X ≤ x, Y < ∞]
∑ P[ X ≤ x , Y = y ], for discrete random var iables
 y
i .e ., FX ( x ) =  x ∞
 
 ∫  ∫ f XY ( x , y )dy dx , for continuous random var iables
 − ∞ − ∞ 
and the marginal distribution of Y with respect to the joint CDF
FXY ( x , y ) is FY ( y ) = P[Y ≤ y] = P[ X < ∞, Y ≤ y ]
∑ P[ X = x , Y ≤ y ], for discrete random var iables

 x
i .e ., FY ( y ) =  y  ∞ 
 ∫  ∫ f XY ( x , y )dx dy , for continuous random var iables
 − ∞ − ∞ 


www.rejinpaul.com
www.rejinpaul.com
COVARIANCE
AND
CORRELATION COEFFICIENT


www.rejinpaul.com
www.rejinpaul.com
COVARIANCE
COVARIANCE(Definition)
The covariance of X and Y, which is denoted by Cov( X, Y)
or σ XY , is defined by Cov( X, Y) = E{[X - E(X)][Y - E(Y)]}
= E[ XY] − E[ X]E[Y]
Note :
(i ) If Cov ( X ,Y ) = 0, we define the two random variables
to be uncorrelated.
(ii) Cov(aX + b, cY + d) = ac Cov(X, Y)


www.rejinpaul.com
www.rejinpaul.com
Result 1 :
Let X and Y be any two random variables and a, b be constants.
Prove that Cov (a X , bY ) = ab cov ( X ,Y )
Result 2 :
Let X and Y be any two random variables and a,b be constants.
Prove that Cov(X + a, Y + b) = Cov(X, Y)
Result 3 :
Show that Cov 2 ( X ,Y ) ≤ Var ( X ).Var (Y )


www.rejinpaul.com
www.rejinpaul.com
CORRELATION COEFFICIENT
CORRELATION COEFFICIEN T (Formula)

We define the correlation coefficient of X and Y, denoted by
ρ ( X ,Y ) or ρ XY , as follows :
Cov ( X , Y ) σ XY
ρ XY = = , − 1 ≤ ρ XY ≤ 1
Var ( X )Var (Y ) σ X σ Y


www.rejinpaul.com
www.rejinpaul.com
Result 1 :
1. Prove that the limits of correlatio n coefficien t is between − 1 and 1.
i.e., − 1 ≤ ρ XY ≤ 1 or ρ XY ≤ 1 or C XY ≤ ρ X ρ Y .
Result 2 :
Prove that two independen t variables are uncorrelated.
Result 3 :
Prove that correlation coefficient is independent of change of
origin and scale.


www.rejinpaul.com
www.rejinpaul.com
Example :
If X 1 has mean 4 and variance 9 while X 2 has mean − 2
and variance 5 and the two variables independen t, find
Var (2 X 1 + X 2 − 5 ).
Solution :
Given E [ X 1 ] = 4 ,Var [ X 1 ] = 9
E [ X 2 ] = − 2 ,Var [ X 2 ] = 5
Var (2 X 1 + X 2 − 5 ) = 4Var X 1 + Var X 2

= 4 (9 ) + 5 = 36 + 5 = 41.


www.rejinpaul.com
www.rejinpaul.com
Example :
The joint p.m.f of X and Y is given below :
X
p(x,y)
-1 1
Find the correlation coefficient of (X,Y).
0 1/8 3/8
Y
1 2/8 2/8
Solution :
The marginal p.m.f of X and Y are given by
X
p(x,y) p(Y= y)
-1 1
0 1/8 3/8 4/8
Y
1 2/8 2/8 4/8
P(X= x) 3/8 5/8 1

www.rejinpaul.com
www.rejinpaul.com
 3  5 2
E[X] = ∑ x i p( x i ) = ( −1)  + (1)  =
i  8  8 8
 3 2 5
E[X ] = ∑ x p( x i ) = ( −1)   + (1)   = 1
2 2
i
2
i  8  8
 4  4 1
E[Y] = ∑ y i p( y i ) = (0)  + (1)  =
i  8  8 2
 4  4 1
E[Y 2 ] = ∑ y i2 p( y i ) = (0) 2   + (1) 2   =
i  8  8 2


www.rejinpaul.com
www.rejinpaul.com
E[XY] = ∑ ∑ x i y j p( x i , y i )
i j
1  3  2  2
= (0)(-1)  + (0)(1)  + (1)(-1)  + (1)(1)  = 0
 8  8  8  8
4 15
σ = E[X ] − [ E ( X )] = 1 − =
2
X
2 2
64 16
15
∴σ X =
4


www.rejinpaul.com
www.rejinpaul.com
1 1 1
σ = E[Y ] − [ E (Y )] = − =
2
Y
2 2
2 4 4
1
∴σ Y =
2
Cov( X, Y )
Hence ρ( X, Y ) =
σ Xσ Y
E[ XY] − E[ X]E[ Y]
=
σ Xσ Y
= −0.2582.


www.rejinpaul.com
www.rejinpaul.com
Example :
Suppose that the two dimensional RVs (X, Y) has the joint p.d.f
 x + y, 0 < x < 1, 0 < y < 1
f(x,y) =  .
0 , otherwise
Obtain the correlation coefficient between X and Y.
Solution :
The marginal density function of X is given by
∞ 1
 y2 
1
1
f X ( x ) = f ( x ) = ∫ f XY ( x , y )dy = ∫ ( x + y )dy =  xy +  = x + .
−∞ 0  2 0 2
Similarly, the marginal density function of Y is given by

∞
1
fY ( y ) = f ( y ) = ∫ f XY ( x , y )dx = y + , 0 < y < 1.
−∞
2


www.rejinpaul.com
www.rejinpaul.com
∞
E[ X] = ∫ xf ( x )dx
−∞
1
 2 x
= ∫  x + dx
0
2
1
 x3 x2  7
= +  = .
 3 4  0 12
∞
E[Y ] = ∫ yf ( y )dy
−∞
1
 2 y
= ∫  y + dy
0
2
1
 y3 y2  7
= +  = .
 3 4  0 12


www.rejinpaul.com
www.rejinpaul.com
E[ X 2 ] = ∫ x 2f ( x )dx
−∞
1
 x x 
1
x 2 4 3
= ∫  x 3 + dx =  + 
0
2  4 6 0
5
= .
12
∞ 1
 y y 
1
y 
2 4 3
E[Y 2 ] = ∫ y 2f ( y )dy = ∫  y 3 + dy =  + 
0
−∞
2  4 6 0
5
= .
12


www.rejinpaul.com
www.rejinpaul.com
∞ ∞
E[ XY] = ∫ ∫ xyf (x, y )dxdy

−∞ −∞
1 1
= ∫ ∫ xy( x + y )dxdy
0 0
1
1 2 
= ∫ y  ∫ ( x + xy )dx dy
0 0 
 y y2 
1
= ∫  + dy
0
3 2
1
= .
3


www.rejinpaul.com
www.rejinpaul.com
5 49 11
Var ( X ) = E[ X ] − [ E ( X )] = −
2 2
=
12 144 144
11
∴σ X = .
12
5 49 11
Var (Y ) = E[Y 2 ] − [ E (Y )]2 = − =
12 144 144
11
∴σ Y = .
12
Coc( X , Y ) E[XY] - E[X]E[Y] 1
Hence ρ (X, Y) = = =− .
σ Xσ Y σ Xσ Y 11


www.rejinpaul.com
www.rejinpaul.com
LINEAR REGRESSION


www.rejinpaul.com
www.rejinpaul.com
LINES OF REGRESSION
The line of regression of y on x is given by

σy
y-y = ρ ( x − x)
σx
The line of regression of x on y is given by

σx
x− x=ρ (y - y )
σy
Note
Both the lines of regression passes through (x, y).


www.rejinpaul.com
www.rejinpaul.com
REGRESSION COEFFICIENTS
σy
Regression coefficien t of y on x is ρ = b yx
σx
σx
Regression coefficien t of x on y is ρ = b xy
σy
PROPERTIES OF REGRESSION
1. Correlation coefficient is the geometric mean between

the regression coefficients. i.e ρ = ± b xy b yx .
2. If one of the regression coefficients is greater than unity,

then other must be less than unity.
3. Regression coefficients are independent of the change of origin

but not of scale.


www.rejinpaul.com
www.rejinpaul.com
Result :
The acute angle between the two lines of regression is
(1− r )σ
σy
2
tan θ =
x
r (σ x + σ y )
2 2
.
Example :
If the equations of the to lines of regression of y on x and x on y
are respective ly, 7x - 16y + 9 = 0; 5y - 4x - 3 = 0, calculate the
coefficien t of correlatio n, x and y .
Solution :
Since both the regression lines passes through (x, y), we get
7x - 16y = -9 & 4x - 5y = -3
3 15
Solving, we get x = - and y = .
29 29


www.rejinpaul.com
www.rejinpaul.com
3 15
∴ The mean value of x and y are - and .
29 29
Now, the regression equation of y onx is,
7x - 16y + 9 = 0
7 9 7
i.e y= x + ⇒ b yx = .
16 16 16
Similarly, the regression equation of x on y is,
5y - 4x - 3 = 0
5 3 5
i.e x= y − ⇒ b xy = .
4 4 4
Hence the correlatio n coefficien t between X and Y is given by

ρ = ± b yx × b xy = ±0.7338 .


www.rejinpaul.com
www.rejinpaul.com
Example
From the following data, find
(i) the two regression equations,
(ii) the coefficient of correlation between the marks in
Economics and Statistics and
(iii) the most likely marks in statistics when marks in
Economics are 30.
Marks in
25 28 35 32 31 36 29 38 34 32
Economics
Marks in
43 46 49 41 36 32 31 30 33 39
Statistics


www.rejinpaul.com
www.rejinpaul.com
Solution :
Let x = marks in Economics & y = marks in Statistics
x−x y−y (x − x) 2 (y − y) (x − x)(y - y )
2
x y
25 43 -7 5 49 25 -35
28 46 -4 8 16 64 -32
35 49 3 11 9 121 33
32 41 0 3 0 9 0
31 36 -1 -2 1 4 2
36 32 4 -6 16 36 -24
29 31 -3 -7 9 49 21
38 30 6 -8 36 64 -48
34 33 2 -5 4 25 -10
32 39 0 1 0 1 0
320 380 0 0 140 398 -93

www.rejinpaul.com
www.rejinpaul.com
where x =
∑ x
= 32; y=
∑ y
= 38
n n
and b YX =
∑ ( x − x )( y − y )
= −0.6643
∑ (x − x) 2
b XY =
∑ ( x − x )( y − y )
= −0.2337
∑ (y − y ) 2
Equation of the line of regression of x on y is

x - x = b XY (y − y )
x = − 0.2337 y + 40 .8806
Equation of the line of regression of y on x is
y − y = b YX (x - x)
y = − 0 .6643 x + 59 .2576


www.rejinpaul.com
www.rejinpaul.com
Hence the correlation coefficient between X and Y is given by

ρ = ± b yx × bxy = ±0.394
When x = 30,
y = (− 0.6643× 30) + 59.2576 = −19.929 + 59.2576 = 39.3286.


www.rejinpaul.com
www.rejinpaul.com
REGRESSION CURVE OR FUNCTION

Regression curve y on x is given by
∞
y = E[Y/X = x] = ∫ y f (y/x )dy

-∞
Regression curve x on y is given by

∞
x = E[X/Y = y] = ∫ x f (x/y )dx

-∞
Note
A linear regression line has an equation of the form
Y = a + bX, where X is the explanator y variable and Y
is the dependent variable. The slope of the line is ' b' and
' a' is the intercept. (the value of Y when x = 0)


www.rejinpaul.com
www.rejinpaul.com
Example :
1, if y = x , 0 < x < 1
Let (X, Y) have the joint p.d.f given by f(x, y) =  .
0, otherwise
Show that the regression of y on x is linear but regression of
x on y is not linear.
Solution :
The marginal p.d.f of X is
∞ x
f(x) = ∫ f(x, y)dy = ∫ dy = 2x

-∞ -x
Similarly, The marginal p.d.f of Y is

∞ 1
f(y) = ∫ f(x, y)dx = ∫ dx = 1

-∞ 0


www.rejinpaul.com
www.rejinpaul.com
Also, the conditional p.d.f of Y is given X = x is given by

f(x, y) 1
f(y/x) = =
f(x) 2x
and the conditional p.d.f of X is given Y = y is given by

f(x, y) 1
f(x/y) = = = 1.
f(y) 1
The regression curve y on x is
∞
y = E[Y/X = x] = ∫ y f (y/x )dy

-∞
x x
1 1  y2  1  x2 x2 
= ∫ y dy =   =  −  = 0
-x
2x 2x  2  − x 2x  2 2


www.rejinpaul.com
www.rejinpaul.com
i.e y = 0 , which is a straight line.

∴ The regression curve y on x is linear.
Regression curve x on y is given by

∞
x = E[X/Y = y] = ∫ x f (x/y )dx

-∞
1 1
 x2  1
= ∫ x dx =   = .
0  2 0 2
Hence the regression curve s on y is not linear.


www.rejinpaul.com
www.rejinpaul.com
TRANSFORMATION OF RANDOVARIABLES


www.rejinpaul.com
www.rejinpaul.com
TRANSFORMATION OF TWO-DIMENSIONAL RANDOM VARIABLE
If (X, Y) is a two - dimensiona l random variable with joint

pdf f XY (x, y) and if U = g(x, y) and V = h(x, y) are two other
random variables then the joint pdf of (U, V) is given by
f UV (u, v ) = f XY (x, y) . J
∂x ∂x
∂ (x, y) ∂u ∂v is called the jacobian of the
where J = =
∂ (u, v) ∂y ∂y
∂u ∂v
transforma tion.


www.rejinpaul.com
www.rejinpaul.com
Result
Assume that U = g(X, Y), and we are required to find the
p.d.f of U. We can use the above transformation method
by defining an auxiliary function W = X or Y so we can
obtain the joint PDF f UW ( u, w ) of U and W.
Then we obtain the required marginal PDF f U ( u) as follows :
∞
f U ( u) = ∫f
−∞
UW ( u, w ) dw
Example :
If X and Y are independen t RVs with pdf' s e - x , x ≥ 0, e - y , y ≥ 0,
X
respectively, find the density functions of U = and V = X + Y.
X+Y
Are U and V independen t?


www.rejinpaul.com
www.rejinpaul.com
Solution :
Since X and Y are independent, f X Y (x y ) = f X (x)fY (y) = e − (x + y) .
x
Solving the equations u = andv = x + y,
x+y
we get x = uv and y = v(1 - u).
∂x ∂x
∂ (x, y) ∂u ∂v = v u
J= = = v.
∂ (u, v) ∂y ∂y − v (1 − u)
∂u ∂v
The joint pdf of (U, V) is given by

f UV (u, v ) = f XY (x, y) . J
= e -(x + y) v = e − v v .


www.rejinpaul.com
www.rejinpaul.com
The range space of (U, V) is obtained as follows :

Since x and y ≥ 0, uv ≥ 0 and 1 - u ≥ 0.
∴ either u ≥ 0 and v ≥ 0 and 1 - u ≥ 0 i.e 0 ≤ u ≤ 1 and v ≥ 0
or u ≤ 0, v ≤ 0 and 1 − u ≤ 0, i.e u ≤ 0 and u ≥ 1, which is absurd.
Therefore range space of (U, V) is given by 0 ≤ u ≤ 1 and v ≥ 0.
∴ f UV (u, v ) = e − v v; 0 ≤ u ≤ 1 and v ≥ 0.
∞
PDF of U is given by f U ( u) = ∫ f UV (u, v ) dv

−∞
∞ ∞
−v −v
= ∫ e − v vdv = (v ) e  − (1) e 

0   −1  1  0
= 1, 0 ≤ u ≤ 1


www.rejinpaul.com
www.rejinpaul.com
PDF of V is given by f V (v ) = ∫ f UV (u, v ) du

−∞
= ∫ e − v vdu
0
= e − v v (1 − 0 ) = e − v v v ≥ 0.
Now f U (u ) × f V ( v ) = (1) × e − v v
= e −v v
= f UV (u, v )
Therefore, U and V are independent RVs.


www.rejinpaul.com
www.rejinpaul.com
Example:
If X and Y are independent RVs each normally distributed with
mean zero and varianceσ 2 , find the density functions of
Y 
R = X2 + Y 2 and θ = tan-1  .
X
Solution :
Since X and Y are independent N(0,σ ),
1 − ( x 2 + y 2 ) / 2σ 2
f XY ( x, y ) = e - ∞ < x, y < ∞ and r = x 2 + y 2 and
2πσ 2
-1  y 
θ = tan   are the transformations from cartesiansto polars.
 x
Therefore,the inverse transformations are given by x = rcosθ and y = rsinθ
∂x ∂x
∂ (x, y) ∂r ∂θ
J= = = r.
∂ (r, θ) ∂ y ∂ y
∂ r ∂θ

www.rejinpaul.com
www.rejinpaul.com
The joint pdf of (R,θ ) is given by

r
f Rθ ( r ,θ ) = e − r 2 / 2σ 2
r ≥ 0, 0 ≤ θ ≤ 2π .
2πσ 2
Since - ∞ < x, y < ∞ and r ≥ 0, 0 ≤ θ ≤ 2π both

represent the entire xy - plane.
The density function of R is given by
2π 2π
r
f R (r ) = ∫ f Rθ (r ,θ )dθ = ∫ e − r 2 / 2σ 2
dθ
0 0
2πσ 2
r
= e − r 2 / 2σ 2
(θ )02π
2πσ 2
r − r 2 / 2σ 2
= ×e r≥0
σ 2


www.rejinpaul.com
www.rejinpaul.com
The density function of θ is given by

∞
r
fθ (θ ) = ∫ e − r 2 / 2σ 2
dr
0
2πσ 2
r2
Putting t = 2 .
2σ
∞
1
∴ fθ (θ ) = ∫
−t
e dt
2π 0
∞
1  e −t  1
=   = , 0 ≤ θ ≤ 2π
2π  − 1  0 2π


www.rejinpaul.com
www.rejinpaul.com


UNIT 2 - Rejinpaul PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

UNIT 2 - Rejinpaul PDF

Hochgeladen von

Copyright:

Verfügbare Formate

www.rejinpaul.

18 December 2014 Two Dimensional Random 1

Download Useful Materials from Rejinpaul.com

MA 6453 PROBABILITY AND QUEUEING THEORY

Joint distributions – Marginal and conditional

18 December 2014 Two Dimensional Random 2

Download Useful Materials from Rejinpaul.com

18 December 2014 Two Dimensional Random 3

Download Useful Materials from Rejinpaul.com

TWO – DIMENSIONAL RANDOM VARIABLES

18 December 2014 Two Dimensional Random 4

Download Useful Materials from Rejinpaul.com

TWO – DIMENSIONAL DISCRETE RANDOM VARIABLES

18 December 2014 Two Dimensional Random 5

Download Useful Materials from Rejinpaul.com

JOINT PROBABILITY MASS FUNCTION (PMF)

When both X and Y are discrete random variables,

18 December 2014 Two Dimensional Random 6

Download Useful Materials from Rejinpaul.com

MARGINAL PROBABILITY DISTRIBUTION

If (X, Y) is a two - dimensional discrete RV, then the

Similarly, the marginal probability mass function of Y, denoted by

18 December 2014 Two Dimensional Random 7

Download Useful Materials from Rejinpaul.com

CONDITIONAL PROBABILITY DISTRIBUTION

18 December 2014 Two Dimensional Random 8

Download Useful Materials from Rejinpaul.com

INDEPENDENT RANDOM VARIABLES

Let (X, Y) be a two - dimensiona l discrete random variable.

18 December 2014 Two Dimensional Random 9

Download Useful Materials from Rejinpaul.com

If X and Y are discrete random variables with the

18 December 2014 Two Dimensional Random 10

Download Useful Materials from Rejinpaul.com

18 December 2014 Two Dimensional Random 11

Download Useful Materials from Rejinpaul.com

Since p( x , y ) is a probability mass function, we have

P [ X = 0] = 7 k + 10k + 13k = 30k =

18 December 2014 Two Dimensional Random 12

Download Useful Materials from Rejinpaul.com

Marginal probabilit y distribution ofY

18 December 2014 Two Dimensional Random 13

Download Useful Materials from Rejinpaul.com

Conditiona l distributi on of X given Y = 1

18 December 2014 Two Dimensional Random 14

Download Useful Materials from Rejinpaul.com

Conditiona l distributi on of X given Y = 2

18 December 2014 Two Dimensional Random 15

Download Useful Materials from Rejinpaul.com

Conditional distribution of X given Y = 3

18 December 2014 Two Dimensional Random 16

Download Useful Materials from Rejinpaul.com

Conditiona l distributi on of Y given X = 0

18 December 2014 Two Dimensional Random 17

Download Useful Materials from Rejinpaul.com

Conditiona l distributi on of Y given X = 1

18 December 2014 Two Dimensional Random 18

Download Useful Materials from Rejinpaul.com

Conditional distribution of Y given X = 2

18 December 2014 Two Dimensional Random 19

Download Useful Materials from Rejinpaul.com

Determine if X and Y are independent.

18 December 2014 Two Dimensional Random 20