Class No.

13,14,15,16,17
Date: August 25,27,28,31, September 01
MA 2302: Introduction to Probability and Statistics

Some Continuous Probability Distributions

Instructor
Prof. Gopal Krishna Panda
Department of Mathematics
NIT Rourkela
Continuous distributions

1. The Uniform or Rectangular Distribution

Let 𝑎 and 𝑏 be two real numbers such that 𝑎 < 𝑏. Then the uniform or rectangular distribution
in the interval [𝑎, 𝑏] has the density function
1
𝑓 𝑥 = ቐ𝑏 − 𝑎 if 𝑎 ≤ 𝑥 ≤ 𝑏
0 otherwise.

If a random variable 𝑋 has the above distribution, it is customary to write 𝑋~𝑈(𝑎, 𝑏).

Example 1. The amount of time of a technician takes to fix the problem of gas leakage of a split AC is
uniformly distributed between 30 minutes to 75 minutes. What percentage of the ACs need at least 40
minutes for the fixation of gas leakage problem? Out of 300 ACs, about how many can be repaired within
40 minutes?
Ans. Let 𝑋 be the amount of time needed to fix the gas leakage problem of an AC. Then
𝑋~𝑈(30,75). Hence,
1/45 if 30 ≤ 𝑥 ≤ 75
𝑓 𝑥 =ቊ
0 otherwise.
Continuous distributions

The probability that a split AC needs at least 40 minutes for the fixation of gas leakage problem
∞ 75
7
= Pr 𝑋 ≥ 40 = න 𝑓(𝑥) 𝑑𝑥 = න 1/45 𝑑𝑥 = = 0.78.
40 40 9

Hence, about 78% ACs need at least 40 minutes for the fixation of gas leakage problem.

The probability that the gas leakage problem if an Ac can be fixed within 45 minutes
45 45
1
= Pr 𝑋 ≤ 45 = න 𝑓(𝑥) 𝑑𝑥 = න 1/45 𝑑𝑥 = .
−∞ 30 3

Thus, out of 300 ACs, about 100 can be repaired within 45 minutes.

Example 2: The amount of time the students of first year B. Tech. take to complete a one km
walk is 7 minutes to 15 minutes. If a first year student is chosen at random, what is the
probability that he will take any amount of time between 10 to 12 minutes? What percentage
of students will take not more than 10 minutes to complete one km walk? What is the
maximum time required by the speediest 10% students?
Continuous distributions
Ans. Let 𝑋 be the amount of time needed by a randomly chosen first year student to walk 1 km.
Then 𝑋~𝑈(7,15). Hence,
1/8 if 7 ≤ 𝑥 ≤ 15
𝑓 𝑥 =ቊ
0 otherwise.

The probability that a randomly chosen 1st year student will take any amount of time between
10 to 12 minutes to walk 1 km
12 12
1
= Pr 10 ≤ 𝑋 ≤ 12 = න 𝑓(𝑥) 𝑑𝑥 = න 1/8 𝑑𝑥 = .
10 10 4

The probability that a student will take not more than 10 minutes to complete one km walk
10 10
3
= Pr 𝑋 ≤ 10 = න 𝑓(𝑥) 𝑑𝑥 = න 1/8 𝑑𝑥 = = 0.375.
−∞ 7 8

Thus, 37.5% of students will not require more than 10 minutes to complete 1 km. Let 𝑡 be the
maximum time required by the speediest 10% students. Then
Pr 𝑋 ≤ 𝑡 = 0.1 ⇒ 𝑡 = 7.8
Continuous distributions
Moments of uniform distribution
Observe that if 𝑋~𝑈(𝑎, 𝑏)
∞ 𝑏 𝑟+1 𝑟+1
1 𝑏 − 𝑎
𝜇𝑟′ = 𝐸 𝑋 𝑟 = න 𝑥 𝑟 𝑓(𝑥) 𝑑𝑥 = න 𝑥 𝑟 ⋅ 𝑑𝑥 = , 𝑟 = 1,2, …
−∞ 𝑎 𝑏−𝑎 (𝑟 + 1)(𝑏 − 𝑎)
Hence,
𝑏 2 − 𝑎2 𝑏+𝑎
𝜇=𝐸 𝑋 = = .
2(𝑏 − 𝑎) 2

𝑏 3 − 𝑎3 𝑏 2 + 𝑏𝑎 + 𝑎2
𝐸 𝑋2 = = .
3(𝑏 − 𝑎) 3
Hence, the variance of the distribution is given by
2 + 𝑏𝑎 + 𝑎2 2 2
𝑏 𝑏+𝑎 𝑏−𝑎
𝜎2 = 𝐸 𝑋2 2
−𝜇 = − = .
3 2 12

Exercise for you: Prove that 𝝁𝟑 = 𝟎. Find 𝝁𝟒 using 𝝁′𝒓 for 𝒓 = 𝟏, 𝟐, 𝟑, 𝟒.

Continuous distributions
Cumulative distribution function of uniform distribution:
If 𝑋~𝑈(𝑎, 𝑏), then 𝐹 𝑥 = 0 if 𝑥 ≤ 𝑎. For 𝑎 ≤ 𝑥 ≤ 𝑏,
𝑥 𝑥
1 𝑥−𝑎
𝐹 𝑥 = න 𝑓 𝑢 𝑑𝑢 = න 𝑑𝑢 = ,
−∞ 𝑎 𝑏−𝑎 𝑏−𝑎
and for 𝑥 ≥ 𝑏, 𝐹 𝑥 = 1. Hence
0 if 𝑥 ≤ 𝑎
𝑥−𝑎
𝐹 𝑥 = i𝑓 𝑎 ≤ 𝑥 ≤ 𝑏
𝑏−𝑎
1 if 𝑥 ≥ 𝑏.
1 𝐹 𝑥 =1
𝑓 𝑥 =
𝑏−𝑎

Total area =1
𝐹 𝑥 =0
𝒂 𝒃 𝒂 𝒃
Sketch of 𝑓(𝑥) Sketch of 𝐹(𝑥)
Continuous distributions
The Exponential distribution

The exponential distribution and the Poisson distribution are very much related in queuing
theory. Under certain assumption, the arrival pattern in a queue (number of arrivals up to a
given point of time) is Poisson while the inter arrival times are exponentially distributed.

A continuous random variable 𝑋 is said to have an exponential distribution with parameter 𝜆 if

its pdf is given by
0 if 𝑥 < 0
𝑓 𝑥 = ቊ −𝜆𝑥
𝜆𝑒 if 𝑥 ≥ 0.
Where 𝜆 > 0. Its cdf is given by 𝐹 𝑥 = 0 if 𝑥 ≤ 0. For 𝑥 ≥0,
𝑥 𝑥
𝐹 𝑥 = න 𝑓 𝑢 𝑑𝑢 = න 𝜆𝑒 −𝜆𝑢 𝑑𝑢 = 1 − 𝑒 −𝜆𝑥 .
−∞ 0
Hence
0 if 𝑥 ≤ 0
𝐹 𝑥 =ቊ
1 − 𝑒 −𝜆𝑥 if 𝑥 ≥ 0.
Continuous distributions
Moments of exponential distribution:

The 𝑟-th moment about origin is given by

∞ ∞
𝜇𝑟′ = 𝐸 𝑋 𝑟 = න 𝑥 𝑟 𝑓(𝑥) 𝑑𝑥 = 𝜆 න 𝑥 𝑟 𝑒 −𝜆𝑥 𝑑𝑥
−∞ 0
For 𝑟 = 1,2 one can easily evaluate the above integral using integration by parts once or twice
and get
1 2
2
𝜇 = 𝐸 𝑋 = and 𝐸 𝑋 = 2
𝜆 𝜆
from which the variance can be obtained as
2 2 2
1
𝜎 =𝐸 𝑋 −𝜇 = 2
𝜆
and consequently, the standard deviation 𝜎 = 1/𝜆. The exponential distribution has the unique
property that its mean and standard deviation are equal.

To calculate the other moments, using repeated integration by parts creates problem. However,
using gamma function, we can avoid complication.
Continuous distributions
The gamma function:

For 𝛼 > 0 is any real number, then we can define an integral

∞
Γ 𝛼 = න 𝑒 −𝑥 𝑥 𝛼−1 𝑑𝑥.
0
Using integration by parts once, one can see that
Γ 𝛼 + 1 = 𝛼Γ 𝛼 .
By direct calculation, one can check that Γ 1 = 1. As an assignment, you verify that
1
Γ = 𝜋.
2
3 1 1 𝜋 5 3 1 1 3 𝜋
Observe that, Γ = Γ = , Γ = ⋅ Γ = ,…. . Moreover, when 𝑛 is a positive
2 2 2 2 2 2 2 2 4
integer,
Γ 𝑛 = 𝑛 −1 Γ 𝑛 −1 = 𝑛 −1 𝑛 − 2 Γ 𝑛 −2 = ⋯ = 𝑛 −1 𝑛 −2 ⋯2⋅ 1 = 𝑛 − 1 !
Continuous distributions
Now, we are in a position to find all moments of exponential distribution. Observe that with a
change of variable 𝑦 = 𝜆𝑥,
∞ ∞ 𝑟
′ 𝑟 𝑟 −𝜆𝑥
𝑦 −𝑦
𝑑𝑦
𝜇𝑟 = 𝐸 𝑋 = 𝜆 න 𝑥 𝑒 𝑑𝑥 = 𝜆 න 𝑒
0 0 𝜆 𝜆
1 ∞ 𝑟+1−1 −𝑦 Γ(𝑟 + 1) 𝑟!
= 𝑟න 𝑦 𝑒 𝑑𝑦 = 𝑟
= 𝑟 , 𝑟 = 1,2, … .
𝜆 0 𝜆 𝜆
Hence,
1 ′ 2 ′ 6 ′ 24
𝜇1′ = , 𝜇2 = 2 , 𝜇3 = 3 , 𝜇4 = 4 .
𝜆 𝜆 𝜆 𝜆
Thus,
1 2 ′ ′2
2 1 1
𝜇= 𝜇1′
= , 𝜎 = 𝜇2 − 𝜇1 = 2 − 2 = 2
𝜆 𝜆 𝜆 𝜆
′ ′ ′ ′3
2 ′ ′ ′ ′ ′2 ′
4 9
𝜇3 = 𝜇3 − 3𝜇2 𝜇1 + 2𝜇1 = 3 , 𝜇4 = 𝜇4 − 4𝜇3 𝜇1 + 6𝜇2 𝜇1 − 3𝜇1 = 4
𝜆 𝜆
Hence,
𝜇32 𝜇4
𝛽1 = 3 = 4, 𝛽2 = 2 = 9.
𝜇2 𝜇2
Continuous distributions
Moment generating function:
∞ ∞
𝐺 𝑡 = 𝐸 𝑒 𝑡𝑋 = න 𝑒 𝑡𝑥 𝑓 𝑥 𝑑𝑥 = 𝜆 න 𝑒 𝑡𝑥 𝑒 −𝜆𝑥 𝑑𝑥
−∞ 0
∞ −1
𝜆 𝑡
= 𝜆 න 𝑒− 𝜆−𝑡 𝑥 𝑑𝑥 = = 1−
0 𝜆−𝑡 𝜆
Observe that the MGF exists for 𝑡 < 𝜆. Moreover,
𝑡 𝑡 𝑡2 𝑡3 𝑡4
log 𝐺 𝑡 = − log 1 − = + 2+ 3+ 4+⋯
𝜆 𝜆 𝜆 𝜆 𝜆
Hence,
𝐺′(𝑡) 1 2𝑡 3𝑡 2 4𝑡 3
= + 2 + 3 + 3 +⋯.
𝐺(𝑡) 𝜆 𝜆 𝜆 𝜆
implying that
2 3
1 2𝑡 3𝑡 4𝑡
𝐺′ 𝑡 = + 2 + 3 + 3 + ⋯ 𝐺(𝑡)
𝜆 𝜆 𝜆 𝜆
Continuous distributions
1
Thus, 𝜇 = 𝐺 ′ (0) = . Now,
𝜆
2 2
1 2𝑡 3𝑡 2 6𝑡 12𝑡
𝐺 ′′ 𝑡 = + 2 + 3 + ⋯ 𝐺 ′ 𝑡 + 2 + 3 + 3 + ⋯ 𝐺(𝑡)
𝜆 𝜆 𝜆 𝜆 𝜆 𝜆
Hence

2 ′′
1 1 2 3
𝐸 𝑋 =𝐺 0 = × + 2= 2
𝜆 𝜆 𝜆 𝜆
and so on. Rest you calculate. Indeed,
∞ ∞
𝑡 −1 𝑡 𝑡2 𝑟! 𝑡 𝑟 𝑡𝑟
𝐺 𝑡 = 1− =1+ + 2+⋯ =1+෍ 𝑟 =1+෍ 𝑟 × .
𝜆 𝜆 𝜆 𝜆 𝜆 𝑟!
𝑟=1 𝑟=1

Hence,
𝑡𝑟 𝑟!
𝜇𝑟′=Coefficient of in the power series expansion of 𝐺 𝑡 = 𝑟 , 𝑟 = 1,2, … . Now
𝑟! 𝜆
substituting 𝑟 = 1,2,3,4 one can find the first four moments of exponential distribution.
Continuous distributions

Example 3: The time between successive births in a city hospital is a random variable which is
exponentially distributed with a mean of 6 hours. If a child birth takes place now, what is the
probability that the next birth will occur not before the completion of next 8 hours?
Ans. Let 𝑋 be the gap between successive births. It is given that the distribution of 𝑋 is
1 1
exponential with mean = 6. hence 𝜆 = and the pdf of 𝑋 is given by
𝜆 6
1 −1 𝑥
𝑓 𝑥 = ቐ6 𝑒
6 if 𝑥 > 0

0 otherwise.
Hence, Pr the next birth will occur not before the completion of next 8 hours
8
−6
= Pr 𝑋 > 8 = 1 − Pr 𝑋 ≤ 8 = 1 − 𝐹 8 = 1 − 1 − 𝑒 = 𝑒 −4/3 = 0.2636.
Continuous distributions
Memoryless property of exponential distribution:

If the distribution of 𝑋 is exponential with parameter 𝜆, then

Pr 𝑋 > 𝑥 + 𝑡|𝑋 > 𝑡 = Pr 𝑋 > 𝑥 .

Proof. Observe that for 𝑥 > 0, the CDF of exponential distribution is 𝐹 𝑥 = 1 − 𝑒 −𝜆𝑥 . Hence
Pr 𝑋 > 𝑥 = 1 − 𝐹 𝑥 = 𝑒 −𝜆𝑥 .
Hence,

Pr 𝑋 > 𝑥 + 𝑡, 𝑋 > 𝑡 Pr 𝑋 > 𝑥 + 𝑡

Pr 𝑋 > 𝑥 + 𝑡|𝑋 > 𝑡 = =
Pr 𝑋 > 𝑡 Pr 𝑋 > 𝑡
1 − 𝐹(𝑥 + 𝑡) 𝑒 −𝜆(𝑥+𝑡) −𝜆𝑥 = Pr 𝑋 > 𝑥 .
= = = 𝑒
1 − 𝐹(𝑥) 𝑒 −𝜆𝑡
Continuous distributions

The gamma distribution

We have already studied the integral
∞
Γ 𝛼 = න 𝑒 −𝑥 𝑥 𝛼−1 𝑑𝑥, 𝛼 > 0.
0
Observe that with a change of variable 𝑦 = 𝜆𝑥, we get
∞ ∞
𝑦 𝛼−1 𝑑𝑦 1 ∞ −𝑦 𝛼−1 Γ(𝛼)
න 𝑒 −𝜆𝑥 𝑥 𝛼−1 𝑑𝑥 = න 𝑒 −𝑦 = 𝛼න 𝑒 𝑦 𝑑𝑦 = 𝛼 .
0 0 𝜆 𝜆 𝜆 0 𝜆
Hence, if we substitute
𝜆𝛼 −𝜆𝑥 𝛼−1
𝑒 𝑥 if 𝑥 > 0
𝑓 𝑥 = ൞Γ 𝛼
0 otherwise,
then, one can check that
∞ ∞
න 𝑓(𝑥) 𝑑𝑥 = න 𝑓(𝑥) 𝑑𝑥
−∞ 0

𝑓(𝑥) so defined is known as the pdf of a gamma distribution. Observe that if 𝛼 = 1, then the
gamma distribution reduces to the exponential distribution.
The Gamma Distribution
Moments of gamma distribution:
∞ 𝛼 ∞
𝜆
𝜇𝑟′ = 𝐸 𝑋 𝑟 = න 𝑥 𝑟 𝑓(𝑥) 𝑑𝑥 = න 𝑥 𝑟 𝑒 −𝜆𝑥 𝑥 𝛼−1 𝑑𝑥
−∞ Γ 𝛼 0
∞
𝜆𝛼 −𝜆𝑥 (𝛼+𝑟)−1
𝜆𝛼 Γ 𝛼+𝑟 𝛼 𝛼+1 𝛼+2 ⋯ 𝛼+𝑟−1
= න 𝑒 𝑥 𝑑𝑥 = ⋅ 𝛼+𝑟
= 𝑟
.
Γ 𝛼 0 Γ 𝛼 𝜆 𝜆
In particular,
𝛼 𝛼 𝛼+1 𝛼(𝛼+1)(𝛼+2) 𝛼(𝛼+1)(𝛼+2)(𝛼+3)
𝜇1′ = , 𝜇2′ = , 𝜇3′ = , 𝜇4′ = .
𝜆 𝜆2 𝜆3 𝜆4
Thus,
𝛼 𝛼 𝛼+1 𝛼 2 𝛼
𝜇= 𝜇1′ 2 2 2
= , 𝜇2 = 𝜎 = 𝐸 𝑋 − 𝜇 = 2
− = 2
𝜆 𝜆 𝜆 𝜆
2𝛼 4 3𝛼 2 + 6𝛼
3 2
𝜇3 = 𝜇3′ − 3𝜇2′ 𝜇1′ + 2𝜇1′ = 3 , 𝜇4 = 𝜇4′ − 4𝜇3′ 𝜇1′ + 6𝜇2′ 𝜇1′ − 3𝜇1′ = .
𝜆 𝜆4
Hence,
𝜇32 4 𝜇4 6
𝛽1 = 3 = , 𝛽2 = 2 = 3 + .
𝜇2 𝛼 𝜇2 𝛼
Continuous distributions
Example 4: Find the harmonic mean of gamma distribution.

Ans. The harmonic mean 𝐻 of a distribution is given by

∞ ∞
1 1 1 𝜆𝛼 1 −𝜆𝑥 𝛼−1
=𝐸 =න ⋅ 𝑓 𝑥 𝑑𝑥 = න ⋅𝑒 𝑥 𝑑𝑥
𝐻 𝑋 −∞ 𝑥 Γ 𝛼 0 𝑥
∞
𝜆𝛼 −𝜆𝑥 𝛼−2
𝜆𝛼
Γ 𝛼−1 𝜆
= න 𝑒 𝑥 𝑑𝑥 = × 𝛼−1
=
Γ 𝛼 0 Γ 𝛼 𝜆 𝛼−1
𝛼−1
which exists for 𝛼 > 1. Hence, 𝐻 = .
𝜆

The moment generating function:

∞ 𝛼 ∞
𝜆
𝐺 𝑡 = 𝐸 𝑒 𝑡𝑋 = න 𝑒 𝑡𝑥 𝑓 𝑥 𝑑𝑥 = න 𝑒 𝑡𝑥 𝑒 −𝜆𝑥 𝑥 𝛼−1 𝑑𝑥
−∞ Γ 𝛼 0
∞ 𝛼
𝜆𝛼 𝜆𝛼
Γ 𝛼 𝜆 𝑡 −𝛼
= න 𝑒− 𝜆−𝑡 𝑥 𝑥 𝛼−1 𝑑𝑥 =
× 𝛼 = = 1− .
Γ 𝛼 0 Γ 𝛼 (𝜆 − 𝑡) 𝜆−𝑡 𝜆
Continuous distributions
Thus,
∞ ∞
𝑡 𝑡 𝑟 𝑟! 𝑡𝑟
log 𝐺 𝑡 = −𝛼 log 1 − =𝛼 ෍ 𝑟 =𝛼 ෍ × 𝑟
𝜆 𝜆 𝑟! 𝜆
𝑟=1 𝑟=1
𝑡𝑟
Observe that the MGF exists for 𝑡 < 𝜆. Now, 𝜇𝑟′ =Coefficient of in the power series
𝑟!
𝛼𝑟!
expansion of 𝐺 𝑡 = , 𝑟 = 1,2, … and the central moments can be calculated as usual.
𝜆𝑟

Example 5: Find the mean of 𝑋, where the distribution of 𝑋 is gamma with parameter 𝛼 =
4 and 𝜆 = 3.
34 27
Ans. Here 𝑓 𝑥 = 𝑒 −3𝑥 𝑥 4−1 = × 𝑒 −3𝑥 𝑥 3 if 𝑥 > 0 and 𝑓 𝑥 = 0 otherwise. Hence
4−1 ! 2
∞
27 ∞ 27 ∞
𝐸 𝑋 =න 𝑥𝑓(𝑥) 𝑑𝑥 = න 𝑥𝑒 −3𝑥 𝑥 3 𝑑𝑥 = න 𝑒 −3𝑥 𝑥 3.5 𝑑𝑥
−∞ 2 0 2 0
27 Γ(4.5) 3.5 × 2.5 × 1.5 × 0.5 × Γ(0.5)
= × 4.5 = 1.5
= 0.63 𝜋.
2 3 2×3
Continuous distributions

The normal distribution:

• The normal distribution is a continuous distribution which is widely used in practice.

Many distributions approaches to normal distribution under certain limiting condition.
The pdf of normal distribution is given by
1 1 𝑥−𝜇 2
−2 𝜎
𝑓 𝑥 = 𝑒 , −∞ < 𝑥 < ∞.
2𝜋 𝜎
If 𝑋 is a random variable with the above pdf, it is customary to write 𝑋~𝑁 𝜇, 𝜎 2 .
• Here, the parameter 𝜇 is positive, negative or zero, while 𝜎 > 0. Later on, we will see that 𝜇
is the mean of the distribution, while 𝜎 is the standard deviation.
• If 𝜇 = 0 and 𝜎 = 1, then the corresponding normal distribution is known as the standard
normal distribution. If the distribution of 𝑋 is standard normal, then it pdf is
1 1
−2𝑥 2
𝑓 𝑥 = 𝑒 , −∞ < 𝑥 < ∞.
2𝜋
Thus, in this case 𝑋~𝑁 0,1 . The standard normal distribution plays a very crucial role while
dealing with any normal distribution.
Continuous distributions
The normal distribution:

• The curve of a normal distribution is perfectly bell shaped and is symmetric about 𝑥 = 𝜇.
𝟏 𝟏 𝒙−𝝁 𝟐
−
𝒇 𝒙 = 𝒆 𝟐 𝝈
𝟐𝝅 𝝈

95.5% 68% 99.7%

𝝁 − 𝟐𝝈 𝝁 𝝁 + 𝟐𝝈 𝝁−𝝈 𝝁 𝝁+𝝈 𝝁 − 𝟑𝝈 𝝁 𝝁 + 𝟑𝝈

• This shows that, though, for a normal distribution, the total area 1 is covered between −∞ to ∞,
one can see that there is hardly any are beyond 𝜇 − 3𝜎 to 𝜇 + 3𝜎.
• Mode is the value of 𝑥 for which 𝑓 𝑥 is maximum. Using the pdf, one can check that the mode
1
of the distribution is equal to 𝜇 where, maximum of 𝑓(𝑥) is . Verify this using calculus.
2𝜋 𝜎
Continuous distributions
The normal distribution:

• Because of symmetry, the area towards left of 𝑥 = 𝜇 is equal to the area towards its right.
That is
𝜇 ∞
න 𝑓(𝑥) 𝑑𝑥 = න 𝑓(𝑥) 𝑑𝑥.
−∞ 𝜇
For example, with 𝑢 = 𝜇 − 𝑥,
𝜇 0 1 ∞ 1
1 − 2 1 − 2
න 𝑓(𝑥) 𝑑𝑥 = − න 𝑒 2𝑢 𝑑𝑢 = න 𝑒 2𝑢 𝑑𝑢
−∞ 2𝜋 𝜎 ∞ 2𝜋 𝜎 0

and with 𝑢 = 𝑥 − 𝜇, one can check that

∞ ∞ 1
1 − 2
න 𝑓(𝑥) 𝑑𝑥 = න 𝑒 2𝑢 𝑑𝑢.
𝜇 2𝜋 𝜎 0
Hence,
𝜇
1
න 𝑓(𝑥) 𝑑𝑥 =
−∞ 2
so that 𝜇 is also median of the distribution.
Continuous distributions
We will show that the mean of the distribution is also equal to 𝜇. Thus, for a normal
distribution, the mean, the median and the mode coincide.

Distribution function: The CDF of normal distribution is

𝑥 𝑥 1 𝑢−𝜇 2
1 −
𝐹 𝑥 = Pr 𝑋 ≤ 𝑥 = න 𝑓(𝑢) 𝑑𝑢 = න 𝑒 2 𝜎 𝑑𝑢,
−∞ 2𝜋 𝜎 −∞

While that for the standard normal distribution is denoted by

𝑥
1 1
−2𝑢2
Φ 𝑥 = න 𝑒 𝑑𝑢.
2𝜋 −∞
This function is very widely used in statistics. For example, if 𝑋~𝑁 𝜇, 𝜎 2 , then with a change
𝑢−𝜇
of variable 𝑧 = , i.e. 𝑢 = 𝜇 + 𝜎𝑧 and 𝑑𝑢 = 𝜎𝑑𝑧, then
𝜎
𝑥−𝜇
𝑥 1 𝑢−𝜇 2
1 − 1 𝜎 1
−2𝑧 2
𝑭 𝒙 = න 𝑒 2 𝜎 𝑑𝑢 = න 𝑒 𝜎𝑑𝑧
2𝜋 𝜎 −∞ 2𝜋 𝜎 −∞
𝑥−𝜇
1 𝜎 1
−2𝑧 2 𝒙−𝝁
= න 𝑒 𝑑𝑧 = 𝚽 .
2𝜋 −∞ 𝝈
Continuous distributions
The relation
𝑥−𝜇
𝐹 𝑥 =Φ
𝜎
is very important in finding probability for normal distribution. In particular, if 𝑋~𝑁 𝜇, 𝜎 2 ,
then
𝑏−𝜇 𝑎−𝜇
Pr 𝑎 < 𝑋 ≤ 𝑏 = 𝐹 𝑏 − 𝐹 𝑎 = Φ −Φ
𝜎 𝜎
which helps in finding the area under the normal curve from 𝑥 = 𝑎 to 𝑥 = 𝑏.
𝟏 𝟏 𝒙−𝝁 𝟐 1 1 2
−
𝒇 𝒙 = 𝒆 𝟐 𝝈 𝑓 𝑥 = 𝑒 −2𝑥
𝟐𝝅 𝝈 2𝜋

𝑏−𝜇 𝑎−𝜇
𝐹 𝑏 −𝐹 𝑎 Φ
𝜎
−Φ
𝜎

𝒂 𝝁 𝒃 𝒂−𝝁 𝟎 𝒃−𝝁
𝝈 𝝈
Continuous distributions
In table 𝐴7, Φ(𝑧) is given for only positive values 𝑧. Observe that Φ 0 = 0.5 and using
−𝑧
1 1
− 𝑥2
Φ −𝑧 = න 𝑒 2 𝑑𝑥
2𝜋 −∞

and with a change of variable 𝑦 = −𝑥, one can see that

−𝑧 ∞
1 1
−2𝑥 2 1 1
−2𝑦 2
Φ −𝑧 = න 𝑒 𝑑𝑥 = න 𝑒 𝑑𝑦
2𝜋 −∞ 2𝜋 𝑧

Using the fact that

∞ 𝑧 ∞
1 1
− 𝑦2 1 1
− 𝑦2 1 1
− 𝑦2
න 𝑒 2 𝑑𝑦 = න 𝑒 2 𝑑𝑦 + න 𝑒 2 𝑑𝑦 = 1
2𝜋 −∞ 2𝜋 −∞ 2𝜋 𝑧
one can see that
𝑧
1 1
−2𝑦 2
Φ −𝑧 = 1 − න 𝑒 𝑑𝑦 = 1 − Φ 𝑧 .
2𝜋 −∞
Continuous distributions
Example 5: If 𝑋~𝑁(50, 225) find (a) Pr 40 ≤ 𝑋 ≤ 70 (b) Pr{ 𝑋 > 45} (c) Pr{ 𝑋 ≤ 55}. Also
(d) find 𝑐 such that Pr 𝑋 > 𝑐 = 0.35 and (e) find 𝑑 such that Pr 41 < 𝑋 < 𝑑 = 0.40.

Ans. (a) 𝑋~𝑁 50, 225 ⇒ 𝜇 = 50, 𝜎 2 = 225 ⇒ 𝜎 = 15. Hence, (using Table A7)
70 − 50 40 − 50
Pr 40 ≤ 𝑋 ≤ 70 = Φ −Φ = Φ 1.33 − Φ −0.67
15 15
= Φ 1.33 − 1 − Φ 0.67 = 0.9082 − 1 + 0.7486 = 0.6568.

(b) Again using Table A7,

45 − 50
Pr{ 𝑋 > 45} = 1 − Pr 𝑋 ≤ 45 = 1 − Φ
15
= 1 − Φ −0.33 = Φ .33 = 0.6293.
(c) Again using Table A7,
55 − 50
Pr{ 𝑋 ≤ 55} = Φ = Φ 0.33 = 0.6293.
15
Continuous distributions
(d) We have to find 𝑐 such that Pr 𝑋 > 𝑐 = 0.35. Hence,
𝑐 − 50
1 − Pr 𝑋 ≤ 𝑐 = 0.35 ⇒ Pr 𝑋 ≤ 𝑐 = 0.65 ⇒ Φ = 0.65
15
From Table A8,
𝑐 − 50
= 0.385. ⇒ 𝑐 = 50 + 15 × 0.385 = 55.775
15

(e) We have to find 𝑑 such that Pr 41 < 𝑋 < 𝑑 = 0.40.

𝑑 − 50 41 − 50 𝑑 − 50 𝑑 − 50
⇒Φ −Φ =Φ − Φ −0.6 = Φ − 1 + Φ 0.6
15 15 15 15
𝑑 − 50
=Φ − 1 + 0.7257 = 0.40.
15
𝑑 − 50 𝑑 − 50
⇒Φ = 0.6743 ⇒ = 0.45
15 15

by virtue table A8. Hence, 𝑑 = 56.75.

Continuous distributions
Example 6: The local authorities in a certain city decide to install 2,000 LED street
lights in a street of the city. If the life time of a light is normally distributed with an
average life of 10,000 hours and standard deviation of 2,000 hours, (a) about how
many lights are expected to fail in the-first 7,000 hours of operation? (b) After what
time 10% of the lights will fail? (c) After what time 90% lights would have been
replaced?
Ans. Let 𝑋 be the lifetime of a random LED street light. Then 𝑋~𝑁(10000, 20002 ).
Probability that an LED light will fail in the first 7,000 hours of operation
7000 − 10000
Pr 𝑋 ≤ 7000 = Φ = Φ −1.5 = 1 − Φ 1.5 = 1 − 0.9332
2000
= 0.0668.
Hence, expected number of failure in the first 7,000 hours of operation is
2000 × 0.0668 ≈ 134.

(b) Pr 𝑋 < 𝑡 = 0.10 , find 𝑡. (c) (b) Pr 𝑋 < 𝑡′ = 0.90, find 𝑡′.
Continuous distributions
Example 7: The monthly incomes of a group of 10,000 persons were found to be normally
distributed with mean ₹ 7500 and standard deviation Rs. 500. Show that, of this group, about
95% had income exceeding ₹ 6680 and only 5% had income exceeding ₹ 8320. What was the
lowest income among the richest 1000?
Ans. Let 𝑋 be the income of a random person. Then 𝑋~𝑁(7500, 5002 ). The probability that
the income of a random person exceeds ₹ 6680 is
6680 − 7500
Pr 𝑋 > 6680 = 1 − Pr 𝑋 ≤ 6680 = 1 − Φ
500
= 1 − Φ −1.64 = Φ 1.64 = 0.9495
Thus, about 95% had income exceeding ₹ 6680. Similarly, the probability that the income of a
random person exceeds ₹ 8320 is
8320 − 7500
Pr 𝑋 > 8320 = 1 − Pr 𝑋 ≤ 8320 = 1 − Φ
500
= 1 − Φ 1.64 = 1 − 0.9495 = 0.0505
Hence about 5% had income exceeding ₹ 8320.
Continuous distributions
Example 7 (continuing): What was the lowest income among the richest 1000?
Let 𝐿 be the lowest income of richest 1000. Then,
1000
Pr 𝑋 ≥ 𝐿 = = 0.1
10000
𝐿 − 7500
⇒1−Φ = 0.1
500
𝐿 − 7500
⇒Φ = 0.9
500
10%
𝐿 − 7500
⇒ = 1.282 (from Table A8) 7500 𝐿
500
⇒ 𝐿 = 7500 + 500 × 1.282 = 8141
Hence, the lowest income of richest 1000 persons is ₹ 8141.
Continuous distributions
Exercise 8: In an university examination of a particular year, 60% of the students failed when
mean of the marks was 50% and standard deviation was 15%. What was the pass mark?
University decided to relax the conditions of passing by lowering the pass marks, to show its
pass result 70%. Find the minimum marks for a student to pass, supposing the marks to be
normally distributed and no change in the performance of students takes place.
Ans. Let 𝐾 be the actual pass mark and 𝑋 be the mark of a random student. Then
𝑋~𝑁(50, 152 ). A student fails if his mark is < 𝐾. As per the data given with the question,
𝐾 − 50 𝐾 − 50
Pr 𝑋 < 𝐾 = 0.60 ⇒ Φ = 0.60 ⇒ = 0.202 ⇒ 𝐾 = 53.03 ≈ 53.
15 15
Let 𝐾 ′ be the revised pass mark so that 70% students can pass. Then
𝐾 ′ − 50 𝐾 ′ − 50
Pr 𝑋 ≥ 𝐾′ = 0.70 ⇒ 1 − Φ = 0.70 ⇒ Φ = 0.30
15 15
𝐾 ′ − 50
⇒ = −0.524 ⇒ 𝐾 ′ = 42.14 ≈ 42.
15
So, with the revised pass mark of 42%, the pass percentage will be 70.
Continuous distributions
Exercise 9: In NIT Rourkela, a student passes in a subject if his total mark is 35 or more. He
gets an Ex grade if he secures 90 or more, A grade if he secures 80 or more but less than 90 and
so on. In the year 2019, 5% students got Ex grade in MA 2203 while 9% student failed.
Assuming a normal distribution of marks, what was the mean mark of MA 2203 in the year
2019 and what was the standard deviation? What percentage of students secured B grade?
Ans. Let 𝑋 be the mark secured in MA 2203 by a random student in the year 2019. Let 𝜇 be the
mean mark and 𝜎 be the standard deviation of marks. Then 𝑋~𝑁(𝜇, 𝜎 2 ). A student fails if his
mark is < 35. As per the data given with the question,
35 − 𝜇 35 − 𝜇
Pr 𝑋 < 35 = 0.09 ⇒ Φ = 0.09 ⇒ = −1.341 ⋯ ⋯ (1).
𝜎 𝜎
90 − 𝜇
Pr 𝑋 ≥ 90 = 0.05 ⇒ 1 − Φ = 0.05
𝜎
90 − 𝜇 90 − 𝜇
⇒Φ = 0.95 ⇒ = 1.645 ⋯ ⋯ (2).
𝜎 𝜎
Continuous distributions
Exercise 9: Continued… What percentage of students secured B grade?
Ans. continued….
From Equations (1) and (2),
90 − 𝜇
= −1. 2267 ⇒ 90 − 𝜇 = 1.2267𝜇 − 42.9695
35 − 𝜇
⇒ 2.2267𝜇 = 132.9695 ⇒ 𝜇 = 59.7159 ≈ 60
and substituting the value of 𝜇 in Equation (2), we get
90 − 𝜇
𝜎= = 18.4098.
1.645
Probability that a random student gets a B grade is
80 − 59.72 70 − 59.72
Pr 70 ≤ 𝑋 < 80 = Φ −Φ
18.41 18.41
= Φ 1.10 − Φ 0.56 = 0.8643 − 0.7123 = 0.1520.
Thus, about 15% students have got B grade.
Continuous distributions
Exercise 10: Steel rods are manufactured to be 3 inches in diameter but they are acceptable if
they are inside the limits 2·99 inches and 3·01 inches. It is observed that 5% are rejected as
oversize and 5% are rejected as undersize. Assuming that the diameters are normally
distributed. find the mean and standard deviation of the distribution. Hence calculate the
proportion of rejections if the permissible limits were widened to 2·985 and 3·015 inches.
You do it.
Continuous distributions

Moments of normal distribution:

Let 𝑋~𝑁(𝜇, 𝜎 2 ). Then the mean is given by

∞ ∞ 1 𝑥−𝜇 2
1 −
න 𝑥 𝑓 𝑥 𝑑𝑥 = න 𝑥 𝑒 2 𝜎 𝑑𝑥
−∞ 2𝜋 𝜎 −∞ 𝑥−𝜇
𝑧=
∞ 𝜎
1 1
−2𝑧 2 ⇒ 𝑥 = 𝜇 + 𝜎𝑧
= න (𝜇 + 𝜎𝑧) 𝑒 𝜎𝑑𝑧 ⇒ 𝑑𝑥 = 𝜎 𝑑𝑧
2𝜋 𝜎 −∞
∞ ∞
𝟏 𝟏
− 𝒛𝟐 𝜎 𝟏
− 𝒛𝟐
=𝜇⋅ න 𝒆 𝟐 𝒅𝒛 + න 𝒛𝒆 𝟐 𝒅𝒛
𝟐𝝅 −∞ 2𝜋 −∞
𝜎
=𝜇×𝟏+ × 𝟎 = 𝜇,
2𝜋
1
1 ∞ −2𝑧 2
since ‫𝑒 ׬‬
2𝜋 −∞
𝑑𝑧 is the total area under the standard normal distribution which is equal
1
∞ − 𝑧2
to 1 and in ‫׬‬−∞ 𝑧𝑒 2 𝑑𝑧, the integrand is an odd function of 𝑧, the value of the integral is 0.
Continuous distributions

Now, the odd ordered central moments are given by

∞
𝜇2𝑟−1 = න 𝑥−𝜇 2𝑟−1 𝑓 𝑥 𝑑𝑥
−∞
∞ 1 𝑥−𝜇 2
1 2𝑟−1 𝑒 −2 𝜎
= න 𝑥−𝜇 𝑑𝑥
2𝜋 𝜎 −∞
∞ ∞
1 1 2
2𝑟−1 𝑒 2𝑧
− 𝜎 2𝑟−1 1 2
= න 𝜎𝑧 𝑑𝑧 = න 𝑧 2𝑟−1 𝑒 2𝑧
−
𝑑𝑧 = 0.
2𝜋 −∞ 2𝜋 −∞

Hence, 𝜇2𝑟−1 = 0 for 𝑟 = 1,2, … .

Continuous distributions
The even ordered central moments are given by
∞
𝜇2𝑟 = න 𝑥−𝜇 2𝑟 𝑓 𝑥 𝑑𝑥
−∞
∞ 1 𝑥−𝜇 2
1 2𝑟 𝑒 −2 𝜎
= න 𝑥−𝜇 𝑑𝑥
2𝜋 𝜎 −∞
∞ ∞
1 1 2
2𝑟 𝑒 −2𝑧
𝜎 2𝑟 2𝑟
1
− 𝑧2
= න 𝜎𝑧 𝑑𝑧 = න 𝑧 𝑒 2 𝑑𝑧
2𝜋 −∞ 2𝜋 −∞

∞ ∞ 1
2𝜎 2𝑟 2𝑟
1
−2𝑧 2 2𝜎 2𝑟 𝑑𝑢 𝑢 = 𝑧2
2
= න 𝑧 𝑒 𝑑𝑧 = න (2𝑢)𝑟 𝑒 −𝑢 ⇒ 𝑑𝑢 = 𝑧𝑑𝑧
2𝜋 0 2𝜋 0 2𝑢 𝑑𝑢
⇒ 𝑑𝑧 =
2𝑢
2𝑟 𝜎 2𝑟 ∞ 𝑟−1 −𝑢 2𝑟 𝜎 2𝑟 1
= න 𝑢 𝑒 𝑑𝑢 =
2 ×Γ 𝑟+
𝜋 0 𝜋 2

𝑟 = 1,2, … .
Continuous distributions
In particular,
Now,
1 1 1 1 3 3
Γ 𝑟+ = 𝑟− Γ 𝑟− = 𝑟− 𝑟− Γ 𝑟−
2 2 2 2 2 2

1 3 3 1 1
= ⋯⋯ = 𝑟 − 𝑟− ⋯ ⋅ Γ
2 2 2 2 2

1 ⋅ 3 ⋅ 5 ⋅ ⋯ ⋅ (2𝑟 − 1) 2𝑟 !
= 𝑟
= 2𝑟 𝜋,
2 2 𝑟!
Hence,
2𝑟 𝜎 2𝑟 1 2𝑟 𝜎 2𝑟 2𝑟 !
𝜇2𝑟 = ×Γ 𝑟+ = × 2𝑟 𝜋
𝜋 2 𝜋 2 𝑟!
2𝑟 ! 2𝑟
= 𝑟 ⋅ 𝜎 , 𝑟 = 1,2, …
2 𝑟!
Continuous distributions
Moments of normal distribution:
2𝑟 ! 2𝑟
𝜇2𝑟−1 = 0, 𝜇2𝑟 = 𝑟 ⋅ 𝜎 , 𝑟 = 1,2, …
2 𝑟!
implies
2!
𝜇2 = 1 ⋅ 𝜎 2 = 𝜎 2 ,
2 1!
𝜇3 = 0,
4!
𝜇4 = 2 ⋅ 𝜎 4 = 3𝜎 4 .
2 2!
Now,
𝜇32 𝜇4
𝛽1 = 3 = 0, 𝛽2 = 2 = 3.
𝜇2 𝜇2
Continuous distributions
Moment generating function:
∞ ∞ 1 𝑥−𝜇 2
1 −
𝐺 𝑡 = 𝐸 𝑒 𝑡𝑋 = න 𝑒 𝑡𝑥 𝑓 𝑥 𝑑𝑥 = න 𝑒 𝑡𝑥 𝑒 2 𝜎 𝑑𝑥
−∞ 2𝜋 𝜎 −∞
∞ ∞ 𝑥−𝜇
1 𝑡(𝜇+𝜎𝑧)
1
− 𝑧2 1 1
− 𝑧2 𝑧=
= න 𝑒 𝑒 2 𝑑𝑧 = 𝑒 𝑡𝜇 ⋅ න 𝑡𝜎𝑧
𝑒 𝑒 2 𝑑𝑧 𝜎
2𝜋 −∞ 2𝜋 −∞ ⇒ 𝑥 = 𝜇 + 𝜎𝑧
∞ ⇒ 𝑑𝑥 = 𝜎 𝑑𝑧
𝑡𝜇
1 1
− 𝑧 2 −2𝜎𝑡𝑧
=𝑒 ⋅ න 𝑒 2 𝑑𝑧
2𝜋
−∞
∞
1 1 1
−2 𝑧−𝜎𝑡 2 +2𝜎2 𝑡 2
= 𝑒 𝑡𝜇 ⋅ න 𝑒 𝑑𝑧
2𝜋 −∞
∞
1
𝜇𝑡+ 𝜎2 𝑡 2 𝟏 𝟏
− 𝒛−𝝈𝒕 𝟐
1
𝜇𝑡+ 𝜎2 𝑡 2
= 𝑒 2 ⋅ න 𝒆 𝟐 𝒅𝒛 = 𝑒 2
𝟐𝝅 −∞
1 ∞ −1 𝑧 2 −𝜎𝑡
since ‫ 𝑒 ׬‬2
2𝜋 −∞
𝑑𝑧 represents the total area of 𝑁(𝜎𝑡, 1) distribution.
Continuous distributions
Moment generating function:
1
𝜇𝑡+2𝜎2 𝑡 2
𝐺 𝑡 = 𝑒
implies that
1 2 2
log 𝐺 𝑡 = 𝜇𝑡 + 𝜎 𝑡
2
and hence,
𝐺′ 𝑡
= 𝜇 + 𝜎 2 𝑡 ⇒ 𝐺 ′ (𝑡) = 𝜇 + 𝜎 2 𝑡 𝐺 𝑡 .
𝐺 𝑡
⇒ 𝐺 ′′ 𝑡 = 𝜇 + 𝜎 2 𝑡 𝐺 ′ 𝑡 + 𝜎 2 𝐺 𝑡
Hence,
𝜇1′ = 𝐺 ′ 0 = 𝜇, 𝜇2′ = 𝐺 ′′ 0 = 𝜇2 + 𝜎 2 ,
2
from which one can easily obtain 𝜇2 = 𝜇2′ −𝜇1′ = 𝜎 2 . By successive differentiation of 𝐺 𝑡 ,
other moments can be obtained.
Continuous distributions
Exercise 11: Prove that if 𝑋~𝑁(𝜇, 𝜎 2 ), then the quartile deviation, the mean deviation and
the standard deviation are approximately in the ratio 10 ∶ 12 ∶ 15.

Ans. There are three quartiles 𝑄1 , 𝑄2 and 𝑄3 that divide a distribution into four equal parts.
For continuous distributions 𝑄𝑖 has the property that the area up to 𝑄𝑖 under 𝑓(𝑥) is equal
to 𝑖/4, 𝑖 = 1,2,3. Thus, 𝐹 𝑄𝑖 = 𝑖/4. The quartile deviation (also known as semi inter
quartile range) is a measure of dispersion defined by
𝑄3 − 𝑄1
𝑄. 𝐷. = .
2
The mean deviation (from mean) is defined as the arithmetic mean or expectation of the
absolute deviations from mean. For frequency distribution, it is defined as
𝑛
1
𝑀. 𝐷. = ෍ 𝑓𝑖 𝑥𝑖 − 𝑥ҧ .
𝑁
𝑖=1
For continuous distribution
𝑀. 𝐷. = 𝐸 𝑋 − 𝜇 .
Continuous distributions
For a normal distribution
𝑄1 − 𝜇 𝑄3 − 𝜇
𝐹 𝑄1 = 0.25, 𝐹 𝑄3 = 0.75 ⇒ Φ = 0.25, Φ = 0.75.
𝜎 𝜎
In view of Table A8,
𝑄1 − 𝜇 𝑄3 − 𝜇 𝑄3 − 𝑄1 2
= −0.674, = 0.674 ⇒ 𝑄. 𝐷. = = 0.674𝜎 ≈ 𝜎.
𝜎 𝜎 2 3
Now,
∞ ∞ 1 𝑥−𝜇 2
1 −
𝑀. 𝐷. = 𝐸 𝑋 − 𝜇 = න 𝑥 − 𝜇 𝑓 𝑥 𝑑𝑥 = න 𝑥−𝜇 𝑒 2 𝜎 𝑑𝑥
−∞ 2𝜋 𝜎 −∞
𝑥−𝜇 ∞ ∞ ∞
𝑧=
𝜎
1 1
−2𝑧 2 2𝜎 1
−2𝑧 2 2𝜎 𝑑𝑢
⇒ 𝑥 = 𝜇 + 𝜎𝑧
= න 𝜎𝑧 𝑒 𝑑𝑧 = න 𝑧𝑒 𝑑𝑧 = න 2𝑢 𝑒 −𝑢
2𝜋 −∞ 2𝜋 0 2𝜋 0 2𝑢
⇒ 𝑑𝑥 = 𝜎 𝑑𝑧
4 1 2
= 2/𝜋𝜎 ≈ 𝜎. 𝑢= 𝑧
2
Hence, 5 ⇒ 𝑑𝑢 = 𝑧𝑑𝑧
2 4 ⇒ 𝑑𝑧 =
𝑑𝑢
𝑄. 𝐷. : 𝑀. 𝐷. : 𝑆. 𝐷. = 𝜎: 𝜎: 𝜎 = 10: 12: 15 2𝑢
3 5
Continuous distributions
Normal approximation to binomial distribution:

Assume that 𝑋~𝐵(𝑛, 𝑝) and 𝑛 is large (say >30). Let 𝑎 and 𝑏 be integers, 0 ≤ 𝑎 ≤ 𝑏 ≤ 𝑛.
Then by De Moivre-Laplace theorem,
Pr 𝑎 ≤ 𝑋 ≤ 𝑏 = Φ 𝑏 ∗ − Φ 𝑎∗
where,
𝑎 − 𝑛𝑝 − 0.5 ∗ 𝑏 − 𝑛𝑝 + 0.5
𝑎∗ = ,𝑏 = .
𝑛𝑝𝑞 𝑛𝑝𝑞
Somebody may ask the role of ±0.5. The binomial distribution is a discrete distribution,
while the normal distribution is a continuous distribution.

Discrete to Continuous →
𝑯𝒊𝒈𝒉𝒕 = 𝑨𝒓𝒆𝒂
0 1 2 3 4 5 6 0 1 2 3 4 5 6
The probability Pr 𝑋 = 𝑘 has been replaced by Pr 𝑘 − 1/2 ≤ 𝑋 ≤ 𝑘 + 1/2 .
Continuous distributions
Example 12: The germination success rate for papaya seeds is 75%. In a package of 400 seeds,
what is the probability that more than 80% of the seeds germinate? What is the probability that
the number of germinations is anywhere between 290 to 308 seeds?
Ans. Let 𝑋 be the number of seeds germinate in a pack of 400. Since, 75% of the seeds
germinate, 𝑝 =probability that any seed germinate= 0.75. Hence 𝑋~𝐵(400,0.75). Hence,
𝑛𝑝 = 300, 𝑛𝑝𝑞 = 75.
Pr More than 80% germination = Pr 𝑋 > 320 = 1 − Pr 𝑋 ≤ 320
320 − 300 + 0.5
=1−Φ = 1 − Φ 2.37 = 1 − 0.9911 = 0.0089.
75
308 − 300 + 0.5 290 − 300 − 0.5
Pr 290 ≤ 𝑋 ≤ 308 = Φ −Φ
75 75
= Φ 0.98 − Φ −1.21 = Φ 0.98 − 1 + Φ 1.21
= 0.8365 − 1 + 0.8869 = 0.7234.
Continuous distributions
Example 13: How many times a fair coin must be tossed so as to obtain at least 500 heads
with a probability of 0.4? (Use normal approximation of binomial distribution.)

Ans. Let 𝑛 be the number of times a fair coin is tossed and let 𝑋 be the number of heads
obtained. Then 𝑋~𝐵(𝑛, 1/2). Hence,
𝑛𝑝 = 𝑛/2, 𝑛𝑝𝑞 = 𝑛/4.
Pr at least 500 heads = Pr 𝑋 ≥ 500
= 1 − Pr 𝑋 ≤ 499
𝑛
499 − + 0.5
=1−Φ 2 = 0.4
𝑛
4
999 − 𝑛
⇒Φ = 0.6
𝑛
Continuous distributions
999 − 𝑛
⇒ = 0.253 ⇒ 999 − 𝑛 = 0.253 𝑛
𝑛
⇒ 0.253 𝑛 = 999 − 𝑛 ⇒ 0.064𝑛 = 998001 + 𝑛2 − 1998𝑛.
⇒ 𝑛2 − 1998.064𝑛 + 998001 = 0 ⇒ 𝑛 ≈ 1007 or 991.
999−𝑛
If 𝑛 = 1007, then Φ < 0.5 and then
𝑛

Pr 𝑋 ≥ 500 = 1 − Pr 𝑋 ≤ 499 > 0.5

Hence, the coin should be tosses 991 times to ensure at least 500 heads with probability
0.4. Observe that if 𝑛 = 991, then
999 − 𝑛
Φ = Φ 0.25 = 0.60
𝑛
and hence
Pr 𝑋 ≥ 500 = 1 − Pr 𝑋 ≤ 499 = 0.4.
Continuous distributions
Example 14: (Bernoulli law of large numbers) . In an experiment let an event 𝐴 is an event
with probability 𝑝 such that 0 < 𝑝 < 1 and let 𝑋 be the number of times 𝐴 happens in 𝑛
independent trials. Show that for any given 𝜖 > 0
𝑋
lim Pr − 𝑝 ≤ 𝜖 = 1.
𝑛→∞ 𝑛
Proof. Observe that X~𝐵(𝑛, 𝑝). Now,
𝑋 𝑛 𝑋 − 𝑛𝑝 𝑛
Pr − 𝑝 ≤ 𝜖 = Pr −𝑛𝜖 ≤ 𝑋 − 𝑛𝑝 ≤ 𝑛𝜖 = Pr − 𝜖≤ ≤ 𝜖 .
𝑛 𝑝𝑞 𝑛𝑝𝑞 𝑝𝑞

𝑋−𝑛𝑝
Since, by De Moivre Laplace theorem, the distribution of approaches to standard normal
𝑛𝑝𝑞
(that is 𝑁(0,1)) as 𝑛 → ∞, hence

𝑛 𝑋 − 𝑛𝑝 𝑛 𝑛 𝑛
Pr − 𝜖≤ ≤ 𝜖 ~Φ 𝜖 −Φ − 𝜖 → Φ ∞ − Φ −∞ = 1 − 0
𝑝𝑞 𝑛𝑝𝑞 𝑝𝑞 𝑝𝑞 𝑝𝑞
=1
Continuous distributions
Ex13: If the lifetime 𝑋 of a certain kind of automobile battery is normally distributed with
a mean of 5 years and a standard deviation of 1 year, and the manufacturer wishes to
guarantee the battery for 4 years, what percentage of the batteries will he have to replace
under the guarantee?
Hints: 𝑋~𝑁(5, 12 ), Ans. 100 × Pr{𝑋 < 4} .
Ex.14 A manufacturer knows from experience that the resistance of resistors he produces
is normal with mean 150 Ω and standard deviation 5 Ω. What percentage of the resistors
will have resistance between and Between 148 Ω and 152 Ω? Between 140 Ω and 160 Ω?
Hints: 𝑋~𝑁(150, 52 ), Ans. 100 × Pr{148 ≤ 𝑋 ≤ 152}.
Ex 15. The breaking strength 𝑋 [kg] of a certain type of plastic block is normally distributed
with a mean of 1500 kg and a standard deviation of 50 kg. What is the maximum load such
that we can expect no more than 5% of the blocks to break?
Hints: 𝑋~𝑁(1500, 502 ), Ans. Find 𝐿 such that Pr 𝐿 ≥ 𝑋 ≤ 0.05 ⇒ 𝐹 𝐿 ≤ 0.05 ⇒ 𝐿?.
Continuous distributions
Ex16: A producer sells electric bulbs in cartons of 1000 bulbs. Find the probability that any
given carton contains not more than 1% defective bulbs, assuming the production process to be
a Bernoulli experiment with 𝑝 = 1% (probability that any given bulb will be defective).

10−10+0.5
Hints: 𝑋= def. bulbs in a cartoon. 𝑋~𝐵(1000, 0.01), Find Pr 𝑋 ≤ 10 = Φ
1000×0.01×0.99

Ex.17 If sick-leave time 𝑋 used by employees of a company in one month is (very roughly)
normal with mean 1000 hours and standard deviation 100 hours, how much time 𝑡 should be
budgeted for sick leave during the next month if 𝑡 is to be exceeded with probability of only
20%?

Hints: 𝑋~𝑁(1000, 1002 ). Find 𝑡 such that Pr{𝑋 > 𝑡} = 0.20.

Ex 18. If the monthly machine repair and maintenance cost 𝑋 in a certain factory is known to be
normal with mean ₹ 120,000 and standard deviation ₹20,000, what is the probability that the
repair cost for the next month will exceed the budgeted amount of ₹150,000?

Hints: 𝑋~𝑁(120000, 200002 ). Find Pr{𝑋 > 150000}.

Continuous distributions
Ex19: If the resistance 𝑋 of certain wires in an electrical network is normal with mean
0.01 Ω and standard deviation 0.001 Ω , how many of 1000 wires will meet the
specification that they have resistance between 0.009 Ω and 0.011 Ω?
0.009−0.01 0.011−0.01
Hints: 𝑋~𝑁 0.01,0.0012 , Pr 0.009 ≤ 𝑋 ≤ 0.011 = Φ −Φ (×1000)
0.001 0.001

Example 20: Prove that

∞ 1 𝑥−𝜇 2 𝑥−𝜇
1 −2 𝜎 𝑧=
න 𝑒 𝑑𝑥 = 1. 𝜎
2𝜋 𝜎 −∞ ⇒ 𝑥 = 𝜇 + 𝜎𝑧
⇒ 𝑑𝑥 = 𝜎 𝑑𝑧
Proof:
∞ 1 𝑥−𝜇 2 ∞
1 − 1 1
−2𝑧 2
න 𝑒 2 𝜎 𝑑𝑥 = න 𝑒 𝑑𝑧 Integrand is an
2𝜋 𝜎 −∞ 2𝜋 −∞ even function of 𝑧.
∞ ∞
2 1
−2𝑧 2 2 𝑑𝑢 1
= න 𝑒 𝑑𝑧 = න 𝑒 −𝑢 𝑢 = 𝑧2
2𝜋 0 2𝜋 0 2𝑢 2
⇒ 𝑑𝑢 = 𝑧𝑑𝑧
1 ∞ −𝑢 1−1 1 1 𝑑𝑢
= න 𝑒 𝑢 𝑑𝑢 =
2 ×Γ = 1. ⇒ 𝑑𝑧 =
𝜋 0 𝜋 2 2𝑢
Continuous distributions
𝑋−𝜇 2
Example 20: If 𝑋~𝑁(𝜇, 𝜎 2 ) and 𝑌 = , then prove that the distribution of 𝑌 is gamma with 𝛼 = λ
𝜎
= 1/2.
Proof: Since If 𝑋~𝑁(𝜇, 𝜎 2 ),
∞ ∞ 1 𝑥−𝜇 2
1 −
න 𝑓(𝑥) 𝑑𝑥 = න 𝑒 2 𝜎 𝑑𝑥 = 1.
−∞ 2𝜋 𝜎 −∞ 𝑥−𝜇
∞ ∞ 𝑧=
1 1
− 𝑧2 2 1
− 𝑧2 𝜎
⇒ න 𝑒 2 𝑑𝑧 = 1 ⇒ න 𝑒 2 𝑑𝑧 = 1 ⇒ 𝑥 = 𝜇 + 𝜎𝑧
2𝜋 −∞ 2𝜋 0 ⇒ 𝑑𝑥 = 𝜎 𝑑𝑧
∞ ∞
2 1
−2𝑦 𝑑𝑦 1 1 1
−2𝑦 2−1
⇒ න 𝑒 =1⇒ න 𝑒 𝑦 𝑑𝑦 = 1 𝑦 = 𝑧2
2𝜋 0 2 𝑦 2𝜋 0 ⇒ 𝑑𝑦 = 2𝑧𝑑𝑧
𝑑𝑦
1 1/2 ∞ ∞
⇒ 𝑑𝑧 =
2 𝑦
1 1
⇒ 2 න 𝑒 −2𝑦 2−1
𝑦 𝑑𝑦 = 1 ⇒ න 𝑔(𝑦) 𝑑𝑦 = 1,
1 0
Γ 0
2
1 1
1/2 1/2 −2𝑦 −1
where 𝑔 𝑦 = 𝑒 𝑦 2 if 𝑦 > 0 and 𝑔 𝑦 = 0 otherwise, which is a gamma distribution.
Γ 1/2