Chapter 6 Continuous Probability Distributions

Section 6.1 Uniform Probability Distribution

General properties of a continuous random variable

1.
probability function, from discrete to continuous:
-- For a discrete r.v., we use the function 𝑃(𝑋 = 𝑥) or 𝑝(𝑥) to denote the probabilities that
r.v. 𝑋 equals different values 𝑥.
-- For a continuous r.v., we need a continuous function, denoted by 𝑓(𝑥), to describe the
relative possibilities at different 𝑥.
For a continuous r.v., we call 𝑓(𝑥) the probability density function (p.d.f.) of 𝑋.
[e.g.] Normal distribution (or the bell-shaped distribution in Chap 3)

-- For a continuous r.v., the probability 𝑋 in an interval is defined as the area under the curve
𝑓(𝑥) in the interval.
Formally speaking,

-- For a continuous r.v., the probability of 𝑋 at a single value 𝑥 is 0, and therefore, 𝑓(𝑥) is
not 𝑃(𝑋 = 𝑥).

2.
The properties of any probability density function 𝑓(𝑥) for a continuous r.v. :
(1) 𝑓(𝑥) should satisfy:
(a) 0 ≤ 𝑓(𝑥), ∀𝑥
(b) ∫ 𝑓(𝑥)𝑑𝑥 = 1 (the area under the curve is 1, since the total probability is 1)
(2) The probability that 𝑋 equals a single value is 0. Therefore,
𝑃(𝑎 ≤ 𝑋 ≤ 𝑏) = 𝑃(𝑎 ≤ 𝑋 < 𝑏) = 𝑃(𝑎 < 𝑋 ≤ 𝑏) = 𝑃(𝑎 < 𝑋 < 𝑏)

A specific continuous random variable: Uniform distribution

3.
[Example] A shuttle bus company recorded the traveling time between Taipei and Taichung.
They found that the traveling time uniformly range between 120 and 140 minutes in
off peak time.
(1) Let 𝑋 denote the time spent from Taipei to Taichung of one shuttle bus

by Y.-W. Chang (NCCU STAT)

 tomorrow. ⇒ 𝑋 is a r.v.
(2) The possible value of 𝑋 is any real number between 120 and 140. ⇒ 𝑋 is a
continuous r.v.

4.
Uniform distribution 𝑋~Uniform (a, b), its probability density function:

𝐸(𝑋) = 𝑉𝑎𝑟(𝑋) =

[e.g.] Continue the example in #3,

[interpretation] For any interval of length 1 minutes in [120, 140], the probability that the
traveling time is within the interval is 0.05.
𝐸(𝑋) = 𝑉𝑎𝑟(𝑋) =

5.
[Example] 𝑋~Uniform (120, 140)

(1) 𝑃(120 ≤ 𝑋 ≤ 130) =

(2) 𝑃(123.75 ≤ 𝑋 ≤ 128.25) =

(3) 𝑃(110 ≤ 𝑋 ≤ 126) =

(4) 𝑃(120 ≤ 𝑋 ≤ 140) =

6.
𝑎+𝑏 (𝑏−𝑎)2
𝑋~Uniform (𝑎, 𝑏). Prove that 𝐸(𝑋) = and 𝑉𝑎𝑟(𝑋) = .
2 12

[proof]

2
Section 6.2 Normal Probability Distribution
7.
[e.g.] the distribution of the heights or weights of a group of people; the distribution of score
of all students participate a test. (higher frequency in the middle and lower frequency
in the two sides)
We use the normal (probability) distribution to depict the distribution for such scenarios.
Figure 6.3

Normal distribution 𝑋~Normal(𝜇, 𝜎 2 ) or 𝑋~𝑁(𝜇, 𝜎 2 ), its probability density function:

, where
𝜇=
𝜎=
𝜎2 =
𝜋 = 3.14159… (PI; the circular ratio)
𝑒 = 2.17828… (Euler's number)

-- 𝜇 and 𝜎 2 are parameters. They determine how the distribution looks like. In particular,
𝜇 determines the location, and 𝜎 2 determines the shape.
-- When 𝜇 = 0 and 𝜎 2 = 1, it is known as the standard Normal distribution. 𝑍~ 𝑁(0, 1).

8.
The properties for the p.d.f. of a Normal distribution:
(1) the maximum of 𝑓(𝑥) occurs at the mean (at 𝑥 = 𝜇)
(2) it is a symmetric function ⇒ mean = median
(3) the 𝑥 −axis is the asymptotic line of 𝑓(𝑥). That is, the tail of 𝑓(𝑥) extend to infinity in
both directions and theoretically never touch the 𝑥 −axis. (𝑓(𝑥) is always larger than 0.)
(4) its skewness is 0

3
(5) its mean, 𝜇, could be positive, 0, or negative (Figure 6.4 in p.288)
(6) its variance, 𝜎 2 , determines the shape of the distribution

9.
As other continuous r.v.’s,

[Proposition 1] (the total area under the curve is 1)

[Proposition 2] (since 𝑓(𝑥) is symmetric about 𝑥 = 𝜇)

In addition, there are three types of quantities that we are interested in:
(1) 𝑃(𝑋 ≤ 𝑎)
(2) 𝑃(𝑎 ≤ 𝑋 ≤ 𝑏)
(3) 𝑃(𝑋 ≥ 𝑎)
where 𝑎 and 𝑏 are constants.

In #10 to #12, we will introduce how to compute (1)~(3) for that 𝑋 is a standard Normal
distribution. In #13, the computation will be generalized to general Normal distribution 𝑋.

10.

𝑃(𝑍 < 𝑧) = 𝑃(𝑍 ≤ 𝑧) =

Compute the probability using the cumulative probabilities table for 𝑍 (p.1070 and 1071 in
the textbook)

(1) 𝑃(𝑍 ≤ −1.65) =

(2) (interpolation) 𝑃(𝑍 ≤ −1.645) = ?

4
(3) 𝑃(𝑍 ≤ 𝑧) is also denoted by 𝐹𝑍 (𝑧) and is the cumulative distribution function for 𝑍.

11.
𝑃(𝑎 ≤ 𝑍 ≤ 𝑏) =

[e.g.1] 𝑃(−0.5 ≤ 𝑍 ≤ 1.25) =

[e.g.2] 𝑃(−1 ≤ 𝑍 ≤ 1) =

The result justifies the statement in Chapter 3 (#23) that, for a bell-shape distribution,
approximately 68.26% of the data whose values will be within 1 standard deviation of
the mean.
[e.g.3] 𝑃(−2 ≤ 𝑍 ≤ 2) =
[e.g.4] 𝑃(−3 ≤ 𝑍 ≤ 3) =

12.
𝑃(𝑍 ≥ 𝑎) =

[e.g.] 𝑃(𝑍 ≥ 1.58) =

13.
For a general Normal distribution 𝑋~𝑁(𝜇, 𝜎 2 ), how to compute the three probabilities in
#9?

Step 1: standardization for each quantity in the inequality

the interpretation for 𝑎(𝑛𝑒𝑤) :

Step 2: compute the probability using the approaches in #10 to #12
[e.g.] 𝑋~𝑁(10, 22 ), compute 𝑃(10 ≤ 𝑋 ≤ 14).

5
14.
For a given probability, find the value of 𝑧:
[e.g.1] Given that 𝑃(𝑍 ≤ 𝑧) =0.1, find the value of 𝑧.

……..

…....

[e.g.2] Given that 𝑃(𝑍 ≥ 𝑧) =0.05, find the value of 𝑧.

15.
[Example]
Grear’s engineering group estimated that the tire mileage: 𝜇 =36500 miles and 𝜎 =5000
miles. The collected data indicate that a normal distribution is a reasonable assumption.

(a) 𝑃(𝑋 ≥ 40000) =

(b) 𝑃(𝑋 ≤ 𝑥) = 0.1, what is the value for 𝑥?

Section 6.3 Normal Approximation of Binomial Probabilities

16.
Some remarks about the Binomial distribution in Section 5.5:
(1) It is not easy to compute its p.m.f. using calculators when 𝑛 is large.
(2) The “shape of the distribution” looks like a Normal distribution when 𝑛 is large.
(Reference: “Chap 6_Demo.pdf” on Moodle)
⇒ When 𝑛 is sufficiently large so that 𝑛𝑝 ≥ 5 and 𝑛(1 − 𝑝) ≥ 5, we would use the
Normal distribution to approximate a Binomial distribution. Specifically, for
𝑋~ Binomial(𝑛, 𝑝), we approximate 𝑋 using 𝑁(𝑛𝑝, 𝑛𝑝(1 − 𝑝)) (≡ 𝑌)
(3) ((2), continued) To compute 𝑃(𝑋 = 𝑥) using 𝑌:
(a) 𝑃(𝑌 = 𝑥) = 0
(b) continuity correction factor:

6
 𝑃(𝑋 = 𝑥) ≈ 𝑃((𝑥 − 0.5) ≤ 𝑌 ≤ (𝑥 + 0.5))

17.
[Example] A company makes errors on 10% of its invoices, according to its historical data.
Taking into consideration of a sample of 100 invoices, what is the probability of
exact 12 invoices with errors? what is the probability of less than or equal to 13
invoices with errors?
[sol] Let 𝑋 denote the number of invoices with error among the 100 invoices.

7
Section 6.4 Exponential Probability Distribution
18.
[e.g. 1] The time between arrivals at a car wash
[e.g. 2] The time between arrivals of customers at a bank
[e.g. 3] The distance between two leaks in 50km of pipeline

Suppose there is one event would occur randomly over a specified interval of time or space
(and satisfy the (a)(b) in #29 in Chapter 5). We usually use the Exponential distribution to
describe the time between two occurrence. In other words, the Exponential distribution is used
to describe the time (or distance) until the next arrival/occurrence.

19.
𝑋~Exponential (𝜆), its probability density function:

where the parameter 𝜆 is the expected value of the r.v.

𝐸(𝑋) = 𝑉𝑎𝑟(𝑋) =

[Example] Suppose the cars arrive at a car wash with the average of 15 minutes between cars.
What is the distribution for the time that we have to wait until the arrival of next car?

(Figure 6.10 + Demo)

20.
𝑋~Exponential (𝜆), its cumulative distribution function:

𝑃(𝑋 ≤ 𝑥) =

[why]

[Example] 𝑋~ 𝑋~Exponential (15)，

𝑃(𝑋 ≤ 6) =

𝑃(6 ≤ 𝑋 ≤ 18) =

8
21.
The relationship between the Poisson distribution and Exponential distribution:
Suppose there is one event would occur randomly over a specified interval of time or space
(and satisfy the (a)(b) in #29 in Chapter 5). We usually use the Exponential distribution to
describe the time between two occurrence. Under the circumstances, the number of
occurrence at a fixed period of time follows a Poisson distribution.

[Example] Suppose we are interested in the number of cars arriving at a car wash. Suppose
further that there are, on average, 5 cars per hour according to historical data. The
number of cars arriving within one hour (denoted by 𝑋 ) follows the Poisson(5)
distribution. On the other hand, since there are, on average, 5 cars per hour, the mean
1
time between arrivals is = 0.2 hr. Thus, the time between any arrivals (denoted by
5

𝑌) follows exponential(0.2) distribution.