Transcript ppt - Department of Mathematics

Chapter 8. Some Approximations to
Probability Distributions: Limit Theorems
Sections 8.4: The Central Limit Theorem
Jiaping Wang
Department of Mathematics
04/22/2013, Monday
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Theorem 8.4
The practical importance of the Central Limit Theorem is that
for large n, the sampling distribution of 𝑋 can be closely
approximated by a normal distribution:
𝑃 𝑋≤𝑏 =𝑃
𝑛 𝑋−𝜇
𝜎
≤
𝑛 𝑏−𝜇
𝜎
= P(Z ≤
𝑛 𝑏−𝜇
𝜎
)
Where Z is the standard normal random variable.
Central Limit Theorem: Let X1, X2, …, Xn be independent and
identically distributed random variables with E(Xi)=μ and
𝑋 −𝜇
1
𝑛
V(Xi)=σ2<∞. Define 𝑌𝑛 = 𝑛( 𝑛 ) where 𝑋𝑛 =
𝑋 .
𝜎
𝑛 𝑖=1 𝑖
Then Yn converges in distribution toward a standard normal
random variable.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
An Application of CLT
Suppose that we wish to find an interval, (a, b), such that
𝑃 𝑎 ≤ 𝑋 ≤ 𝑏 = 0.95 which is equivalent to
𝑛 𝑎−𝜇
𝑛 𝑋−𝜇
𝑛 𝑏−𝜇
𝑃(
≤
≤
)=0.95, approximately,
𝜎
𝜎
𝜎
𝑛 𝑎−𝜇
𝑛 𝑏−𝜇
𝑃(
≤𝑍≤
)=0.95, so we can have
𝜎
𝜎
𝑛 𝑎−𝜇
𝑛 𝑏−𝜇
1.96𝜎
=-1.96 and
=1.96  𝑎 = 𝜇 −
𝜎
𝜎
𝑛
1.96𝜎
And b= 𝜇 +
.
𝑛
it is
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Example 8.6
From 1976 to 2002, a mechanical golfer, Iron Byron, whose swing was modeled
after that of Byron Nelson (a leading golfer in the 1940s), was used to
determine whether golf balls met the Overall Distance Standard. Specifically,
Iron Byron would be used to hit the golf balls. If the average distance of 24
golf balls tested exceeded 296.8 yards, then that brand would be considered
nonconforming. Under these rules, suppose a manufacturer produces a new
golf ball that travels an average distance of 297.5 yards with a standard
deviation of 10 yards.
1. What is the probability that the ball will be determined to be
nonconforming when tested?
2. Find an interval that includes the average overall distance of 24 golf balls
with probability of 0.95.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Solution
Answer:
1. Assume n=24 is large enough for the approximation. The μ=297.5 and
𝜎
10
296.8−297.5
=
=
2.04.
Thus,
𝑃
𝑋
>
296.8
≈
𝑃
𝑍
>
= 𝑃 𝑍 > −0.34
𝑛
24
2.04
= 0.5 + 0.1331 = 0.6331.
2. We have seen that
𝜎
𝜎
𝑃[𝜇 − 1.96( 𝑛) ≤ 𝑋 ≤ 𝜇 − 1.96( 𝑛)]=0.95 
𝜎
𝜇 − 1.96( 𝑛)=297.5-1.96(10/(24)1/2)=293.5 and
𝜎
𝜇 + 1.96( 𝑛)=297.5+1.96(10/(24)1/2)=301.5
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Example 8.7
A certain machine that is used to fill bottles with liquid has been observed over
a long period, and the variance in the amounts of fill has been found to be
approximately σ2=1 ounce. The mean ounces of fill μ, however, depends on an
adjustment that may change from day to day or from operator. If n= 36
observations on ounces of fill dispensed are to be taken on a given day (all with
the same machine setting), find the probability that the sample mean will be
within 0.3 ounce of the true population mean for the setting.
Answer: n=36 is large enough for the approximation.
𝑃 𝑋 − 𝜇 ≤ 0.3 = 𝑃 −0.3 ≤ 𝑋 − 𝜇 ≤ 0.3 = 𝑃 −0.3 6 ≤ 𝑍 ≤ 0.3 6
=P(-1.8≤Z≤1.8)=2(0.4641)=0.9282.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Approximate Binomial by Normal Distribution
Let 𝑋 = 𝑛𝑖=1 𝑌𝑖 with Bernoulli trial Yi having probability p for success.
𝑋
1 𝑛
So =
𝑌𝑖 = 𝑌, then we have E(X/n)=p, V(X/n)=p(1-p)/n, the normality
𝑛
𝑛 𝑖=1
follows from the CLT. As 𝑋 = 𝑛𝑌, X has approximately a normal distribution with
a mean of np and a variance of np(1-p).
𝑃(𝑋 ≤ 𝑎) ≈ 𝑃(𝑍 ≤
𝑎 + 0.5 − 𝑛𝑝
)
𝑛𝑝 1 − 𝑝
If we can make sure 𝑝 ± 2 𝑝(1 − 𝑝)/𝑛 will lie within the interval (0,1) where
0.5 is the correct factor.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Example 8.8
Six percent of the apples in a large shipment are damaged. Before accepting
each shipment, the quality control manager of a large store randomly selects
100 apples. If four or more are damaged, the shipment is rejected. What is the
probability that this shipment is rejected?
Answer: Check 𝑝 ± 2 𝑝(1 − 𝑝)/𝑛 = 0.06 ± 0.024 which is entirely within (0,1). Thus
The normal approximation should work.
𝑃 𝑋 ≥4 =1−𝑃 𝑋 ≤3 ≈1−𝑃 𝑍 ≤
3 + 0.5 − 6
100 0.06 0.94
= 1 − 𝑃 𝑍 < −1.05
=1-(0.5-0.3531)=0.8531.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Example 8.9
Candidate A believe that she can win a city election if she receives at least 55%
of the votes from precinct I. Unknown to the candidate, 50% of the registered
voters in the precinct favor her. If n=100 voters show up to vote at precinct I,
what is the probability that candidate A will receive at least 55% of that
precinct’s votes?
Answer: Let X denote the number of voters in precinct I who vote for candidate A.
The probability p that a randomly selected voter favors A is 0.5, then X can be
Considered as a binomial distribution with p=0.5 and n=100. Approximately by normal
Distribution, we need find
𝑋
𝑃
≥ 0.55 = 1 − 𝑃 𝑋 < 0.55𝑛 = 1 − 𝑃 𝑋 ≤ 54
𝑛
54 + 0.5 − 50
≈1−𝑃 𝑍 ≤
= 1 − 𝑃 𝑍 ≤ 0.9
100 0.5 0.5
=0.5-0.3159=0.1841.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Additional Example 1
Apply the central limit theorem to approximate P[X1+X2+…+X20≤ 50], where
X1, …, X20 are independent random variables having a common mean μ= 2
and a common standard deviation σ= 10.
Answer: P[X1+X2+…+X20≤ 50] = P[(X1+X2+…+X20)/20≤ 50/20
=𝑃 𝑋20 ≤ 2.5 = 𝑃 20(𝑋 20 − 𝜇)/𝜎 ≤ 20(2.5 − 2)/10 ≈ 𝑃(𝑍 ≤ 0.224)=0.5+0.0871
=0.5871.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Additional Example 2
Let X have a binomial distribution Bin(200,0.15). Find the normal approximation
to P[25 <X < 35].
Answer: P(25<X<35)=P(X<35)-P(X≤25)=P(X≤34)-P(X≤25)
34 + 0.5 − 200 0.15
25 + 0.5 − 200 0.15
≈𝑃 𝑍≤
−𝑃 𝑍 ≤
200 0.15 0.85
200 0.15 0.85
=P(Z≤0.89)-P(Z≤-0.89)=0.3133(2)=0.6266.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Additional Example 3
Roll a fair coin 100 times, use CLT to find the approximate probability that more
than 60 tails shows.
Answer: n=100, p=0.5, X=number of tails.
P(X>60)=1-P(X≤60)≈ 1 − 𝑃 𝑍 ≤
60+0.5−100 0.5
100 0.5 0.5
=1-P(Z≤2.1)=0.5-0.4772=0.0822.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL
Homework #11
Page 405: 8.2, 8.4, 8.6 (a);
Page 417: 8.12, 8.16, 8.18, 8.28.
Due next Monday, 04/29/2013.
The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL