chi-square and f distribution
TRANSCRIPT
Distributions Arising from the Normal
Chi-Square and F-Distributions
Instructor: Dr. Gaurav Bhatnagar
22001: PROBABILITY AND STATISTICS
Chi-square Distribution
A random variable is having Chi-square distribution if itcan be written as the sum of squares of mutual independentstandard normal random variables, i.e.,
All
The number of standard normal random variable is calledthe degree of freedom of the random variable.
Notation:
X
2 2 2 21 2 3 ... (1)nX Z Z Z Z
~ (0,1)iZ N
2~ nX
PDF of Chi-square Distribution
Given:
Let be the MGF of X, then
where
2 2 2 21 2 3 ... nX Z Z Z Z
2 22
1
( ) ( ) ( ) ( )i i
n n
X Z ZZ i
M t M t M t M t
( )XM t
2 2
2
21
1 2 /2 2
( ) ( )
1 11 2 ,22
i i
i
i
t z t zi iZ
t zi
M t E e e f z dz
e dz t t
22
1( ) ( ) 1 2 , (2)2i
nn
X ZM t M t t t
PDF of Chi-square Distribution
Recall the MGF of , which is given by
Comparing Eqns. (2) and (3)
~ ( , )Y
( ) (3)YM tt
1( ) ( ) iff and2 2
1~ ,2 2
X YnM t M t
nX
22( ) ( ) 1 2 (2)
i
nn
X ZM t M t t
PDF of Chi-square Distribution
Since
Then the pdf of X is given by
1~ ,2 2nX
12
212 2( ) , 0 (4)
( / 2)
nx xe
f x xn
A random variable has the Chi-square distribution with
n degree of freedom if its pdf is given by Eqn. (4).X
PDF of Chi-square Distribution
Chi-square Distribution
'
12 2
2
( ) ( )
1
2 ( / 2)
r rr
n xr
n
E X x f x dx
x e dxn
Moments:
Mean: r = 1
Variance:
' 2 2
2
1
2 ( / 2)
n x
r n x e dx nn
2' 2 2( ) ( ) ( 2) 2r E X E X n n n n
Chi-square Distribution
(1 2 )12 2
02
2
( ) ( ) ( )
1
2 ( / 2)
1(1 2 ) ,2
xt xtX
n t x
n
n
M t E e e f x dx
x e dxn
t t
Moments Generating Function:
Let and be two independent Chi-square random
variables with degree of freedom n and m, respectively.Then the sum is also a Chi-square random variable with
n+m degree of freedom.
Proof: The proof is straightforward using MGF.
1X
1 2 1 2
/2 /2 ( )/2
21 2
( ) ( ) ( )
(1 2 ) (1 2 ) (1 2 )~
X X X X
n m n m
n m
M t M t M t
t t tX X
Chi-square Distribution
2X
Probabilities with Chi-square distribution: Let be anyvalue between (0,1). Then there exist such that
2,nP X
Chi-square Distribution
2,n
2,n
Find 20.01,21 0.01P X
Chi-square Distribution20.01,21
F Distribution
A random variable is having F distribution if it can bewritten as the ratio of two mutual independent Chi-squarerandom variables, i.e.,
where and
The RV X is said to be of degree of freedom m and n.
Notation:
X
1
2
/ (6)/
Q mXQ n
21 ~ mQ
,~ m nX F
22 ~ nQ
PDF of F Distribution
21
2
2
( ) , 0 (7), 12 2
m
m
m n
mxnf x x
m n mB xn
A random variable has the F distribution with m and ndegree of freedom if its pdf is given by following equation.
X
F Distribution
Mean:
Variance:
Moment Generation Function: Does not exist (Check!!!)
'1 , 2
2n n
n
222
22
2
2
( ) ( )
( 2)( 2)( 4) 2
2 ( 2) , 4( 2) ( 4)
E X E X
n m nm n n n
n n m nm n n
Probabilities with F distribution: Let be any value between(0,1). Then there exist such that
, ,m nP X F
F Distribution
, ,m nF
, ,m nF
Important relation:
1, , , ,
2
2 2
1 , , 1 , ,
2
1 , ,
//
/ /1 11/ /
/ 1 1 (8)/
m n m n
m n m n
m n
Q mP X F P FQ n
Q n Q nP PQ m F Q m F
Q nPQ m F
F Distribution
Important relation (Contd…):
From Eqns. (8) and (9), we have
21 , ,
1
/1 (9)/ m n
Q nP FQ m
F Distribution
1 , ,, ,
1 (10)m nm n
FF
This relation says:
F Distribution
0.9,5,70.1,7,5
1 1 0.29673.37
FF