PSTAT 120B Practice Midterm
Instructions: 1 Closed books and Notes. 2 A cheat/formula sheet allowed and Density and
Distribution Tables are allowed. 3. Calculator is allowed. 4 Answer all questions and write
them clearly/neatly. 4 Need extra space, write answers back. 5. Midterm includes 4 or 5
questions; 6. Dont ignore the bivariate transformations.
1 (This is for practice from 120A class. Dont expect it in the midterm). Let X have the pdf
fX (x) = ex for x > 0 and > 0.
a. Derive the MGF of X.
b. Derive E(X) using MGF.
Answers:
a (7pts):
Z
etx ex dx
MX (t) =
Z0
e(t)x dx
=
0
(t)x
=
e
( t)
0
, for t <
= 0
( t)
MX (t) =
, for t < .
t
b (3pts).
dMX (t)
E(X) =
dt t=0
=
( t)2 t=0
1
E(X) =
.
2 (10pts). Let X have the pdf fX (x) = ex for x > 0 and > 0. Suppose Y = X.
Derive the pdf of Y .
Answers:
dy
y = g(x) = x, dx
= 21 x > 0, for allx > 0. Hence it is increasing function. Therefore X to
Y is one-to-one onto transformation.
x = g 1 (y) = y 2 . Hence
dx
dy
= 2y.
Since the transformation one-to-one onto and increasing,
dy
1
fY (y) = fX (g (y))
dx
fY (y) = 2 y ey
for 0 < y < .
Note: Only for the domain y (0, ), fY (y) integrates to 1.
3 (10pts). Let X1 , , Xn be a random sample from thePpopulation X with the density
n = 1 n Xi .
function fX (x) = ex for x > 0 and > 0. Let X
i=1
n
n.
a. Find the MGF of X
n ).
b. Find the E(X
Answers:
a.
For iid random variables:
n
t
MX ( )
MX n (t) =
n
n
=
nt
n
n
MX n (t) =
.
n t
b.
dMX n (t)
E(X n ) =
dt
t=0
n1
n
n
= n
2
(n t)
(n t)
t=0
n1
n
n
= n
(n)
(n)2
1
E(X n ) =
.
n = 1 Pn Xi and S 2 = 1 Pn (Xi
4 (10pts). Let X1 , , Xn be a random sample and let X
n
i=1
i=1
n
n1
n )2 . Suppose another independent observation Xn+1 becomes available. Show the following
X
recursion equations are true.
n+1 = Xn+1 +nXn
a. X
n+1
n
2
n )2 .
b. n(Sn+1
) = (n 1)Sn2 + n+1
(Xn+1 X
part b for Computer Science students; how to update information from the past
with the present without repeating calculations; dont expect it in the midterm.
Answers:
a. (5pts) 4pts for showing the work and 1 pt for the end result.
Pn+1
Xi
n+1 P
Xn+1 + ni=1 Xi
=
n+1
Xn+1 + nX n
X n+1 =
.
n+1
Pn
P
Xi
Note that X n = i=1
, that gives nX n = ni=1 Xi .
n
X n+1 =
i=1
b. (5pts)
Using the well known result for sample variance Sn2 =
variance of X1 , , Xn , Xn+1 ,
n+1
2
Sn+1
n+1
X
1X
1
=
(Xi X n+1 )2 =
n i=1
n
1
n1
P
n
i=1
Xi2
2
nX n
, the sample
!
Xi2 (n +
2
1)X n+1
i=1
Then,
2
nSn+1
=
n+1
X
!
2
Xi2
(n + 1)X n+1
i=1
n
X
!
2
2
(n + 1)X n+1
+ Xn+1
Xi2
i=1
n
X
!
Xi2
2
nX n + nX n + Xn+1
(n + 1)X n+1 , (adding and subtracting nX n )
i=1
2
2
nSn+1
2
(n + 1)X n+1
= (n 1)Sn2 + nX n + Xn+1
2
Xn+1 + nX n
2
2
2
(using the result from part a)
= (n 1)Sn + nX n + Xn+1 (n + 1)
n+1
(n + 1) 2
2
2
2 2
= (n 1)Sn2 + nX n + Xn+1
X
+
n
X
+
2nX
X
n n+1
n+1
n
(n + 1)2
!
2
2
Xn+1
n2 X n
n
2
2
2
+ Xn+1
2
= (n 1)Sn + nX n
X n Xn+1
n+1
n+1
n+1
n
n
n
2
2
2
= (n 1)Sn +
X +
X
X n Xn+1
2
n + 1 n n + 1 n+1
n+1
n 2
2
2
X n + Xn+1 2X n Xn+1
= (n 1)Sn +
n+1
n
n 2 .
= (n 1)Sn2 +
Xn+1 X
n+1
5 (20 pts).
iid samples from the pdf fX (x) = 6x(1 x), 0 < x < 1. Let
PnLet X1 , 2, Xn be
P
1
1
X n = n i=1 Xi and S = n1 ni=1 (Xi X n )2 .
a. Find E(X n ).
b. Find Var(X n ).
c. Find E(S 2 ).
d. Explain what happens to E(X n ) and Var(X n ) at n = .
Answers:
Notice that fX (x) = 6x(1 x), 0 < x < 1 is Beta distribution with = 2 and = 2. Then
2
1
E(X) = +
= 21 and Var(X) = (+
) ( + + 1) = 20
.
a. Using the properties of identically distributed samples, E(X n ) = E(X) = 21 .
b. Using the properties of iid samples, Var(X n ) =
Var(X)
n
1
.
20n
c. Using the properties of iid samples, E(S 2 ) = Var(X) =
d. E(X n ) =
1
2
1
.
20
stays as it is when n = but Var(X n ) = 0.
6 (15 pts). Let X1 , , Xn be iid samples from the pdf fX (x) =
sample minimum as X(1) .
a. Find the sampling distribution of X(1) .
b. Find the expected value and the variance of X(1) .
c. Explain what happens to E(X(1) ) and Var(X(1) ) at n = .
Answers:
fX (x) =
2
,
x3
1 < x < . Then FX (x) =
Rx
1
fX (u)du = 1
2
,
x3
1
.
x2
a. For iid samples, fX(1) (x) = nfX (x)(1 FX (x))n1 = n x23 (1 1 +
2n
fX(1) (x) = x2n+1
, 1 < x < .
R
R
dx =
b. EX(1) = 1 xfX(1) (x)dx = 1 x2n
R 2
R 2n2n
2
EX(1) = 1 x fX(1) (x)dx = 1 x2n1 dx =
2
VarX(1) = EX(1)
(EX(1) )2 =
n
(n1)
c. At n = , EX(1) 1 and VarX(1) =
2n
.
2n1
2n
n
= (n1)
.
2(n1)
2
2n
= (2n1)n2 (n1) .
2n1
n
(2n1)2 (n1)
n
n3
1 < x < . Denote the
1
n2
= 0.
1 n1
)
x2
2n 1 n1
( )
x3 x2
