AMS 311 (Fall, 2009)
Joe Mitchell
PROBABILITY THEORY
Homework Set # 9 – Solution Notes
(1).
(20 points)
The average number of bank failures in the USA is two per week.
(a). Estimate the probability,
p
, that at least five bank failures occur in the USA next week. (What inequality are you
using?)
Let
X
be the number of bank failures that occur in the USA next week. We know that
E
(
X
) = 2.
We use Markov’s inequality to get a bound on
p
:
0
≤
p
=
P
(
X
≥
5)
≤
E
(
X
)
5
=
2
5
(b).
Assume now (for parts (b), (c), and (d)) that you are told that the variance of the number of bank failures in
the USA in any one week is 4. Now give an improved estimate of
p
(using an inequality).
Now we know that
var
(
X
) = 4. So we apply the onesided Chebyshev inequality, with
μ
=
E
(
X
) = 2 and
a
= 3:
0
≤
P
(
X
≥
5) =
P
(
X
≥
2 + 3)
≤
4
4 + 3
2
=
4
13
(You could have used a 2sided Chebyshev, which would result in a weaker bound as follows:
P
(
X
≥
5)
≤
P
(
X
≥
5 or
X
≤ 
1) =
P
(

X

2

≥
3)
≤
4
3
2
=
4
9
.)
(c).
Give a Central Limit Theorem estimate for the probability
q
that during the academic year (considered to be 31
weeks total, fall and spring) there are more than 75 bank failures in the USA.
Let
Y
i
be the number of bank failures in
week
i
, for
i
= 1
, . . . ,
31. Let
Y
=
∑
31
i
=1
Y
i
denote the total number of bank failures during the academic year. Then,
E
(
Y
) = 31
·
E
(
Y
i
) = 31
·
2 = 62 and
var
(
Y
) = 31
·
var
(
Y
i
) = 124 (using the fact that the
Y
i
’s are independent). Then,
using the Central Limit Theorem estimate, we get
q
=
P
(
Y >
75) =
P
(
Y

62
√
124
>
75

62
√
124
)
≈
P
(
Z >
1
.
17) = 1

Φ
(1
.
17) = 1

.
879 =
.
121
(d). Use an inequality to get the best bounds you can on the probability
q
estimated in part (c).
We can apply the onesided Chebyshev inequality to get
0
≤
P
(
Y >
75) =
P
(
Y >
62 + 13)
≤
124
124 + 13
2
=
.
4232
(Note that Markov inequality gives
P
(
Y >
75)
≤
62
75
, which is weaker than what we get above.)
(2).
(15 points)
The cash register is broken at the bagel shop, so transactions are done by hand and recorded. Your lazy
employee, Joe, charges customers exactly (e.g., if they owe
$
3.61, he makes sure they pay exactly
$
3.61); however, lazy
Joe only records in the notebook the dollar amount of each transaction (e.g.,
$
3 for the
$
3.61 transaction), omitting the
cents. Assume that Joe serves 20 customers today. Find an upper bound on the probability that the total shown in Joe’s
notebook is at least
$
15 less than the actual amount of money those 20 customers paid to Joe.
Let
X
i
be the amount (in dollars) Joe fails to record when he writes the
i
th customer’s purchase in the notebook.
Then,
X
i
is Uniform(0,1).
(Actually,
X
i
is discrete uniform on the set
{
0
,
0
.
01
, . . . ,
0
.
99
}
, but we often approximate
it with a continuous uniform. You can readily compute
E
(
X
i
) =
∑
99
i
=0
i
100
1
100
=
99
200
≈
0
.
5 and
var
(
X
i
) =
E
(
X
2
i
)

[
E
(
X
i
)]
2
=
∑
99
i
=0
(
i
100
)
2
1
100

[
E
(
X
i
)]
2
≈
1
/
12 ) Let
X
=
X
1
+
· · ·
+
X
20
be the total amount (in dollars) that Joe fails to
record for the 20 transactions; then, his notebook ends up showing $
X
dollars less than it should (i.e., less than what
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
This is the end of the preview.
Sign up
to
access the rest of the document.
 Spring '10
 all
 Central Limit Theorem, Variance, Probability theory

Click to edit the document details