Probability Distributions

Recapitulation:
A random variable X is a function from the sample space into the set of real numbers. A random
variable can be discrete or continuous. The set of all possible values of a random variable together
with the corresponding probabilities is called the probability distribution or simply, the
distribution of the random variable.
We will study the following:
1) Discrete Distributions
 Binomial Distribution
 Poisson Distribution
2) Continuous Distributions
 Normal Distribution
 Exponential Distribution
 Gamma Distribution
 Chi-Square Distribution
 Uniform Distribution

Bernoulli Trials
It happens very often in real life that an event may have only two outcomes that matter. For
example, you may get a head or a tail when a coin is tossed, either you pass an exam or you do
not pass an exam, either you get the job you applied for or you do not get the job, either your flight
is delayed or it departs on time, etc. The probability theory abstraction of all such situations is
a Bernoulli trial.
Trials are said to be Bernoulli trials if they satisfy the following:
i) Every trial has only two outcomes, arbitrarily called success (S) and failure (F).
ii) Probability of success, denoted by p, remains constant from trial to trial.
iii) Trials are independent.
Probability of failure is denoted by q, p + q = 1.
Binomial Distribution:
This distribution is based on Bernoulli trials. The assumptions underlying the Binomial
distribution are
 There are only two outcomes – success and failure
 p, the probability of success remains the same for all the trials.
 There are n independent trials
Define the random variable X by
X = number of successes
Then X takes values 0, 1, 2, ….., n. We call X a binomial random variable with parameters n and
p and denote this by XB(n, p).

1
Theorem: Let X be binomial random variable with parameters n and p. Then
Pr{𝑋 = 𝑟} = nCrprqn-r, r = 0, 1, 2, …, n
Proof: X = r means, of the n trials, r trials resulted in success and the remaining n – r resulted in
failure. Consider a particular element of the sample space for which X = r, say
𝑆𝑆 … . . 𝑆 𝐹𝐹 … … 𝐹
𝑟 𝑡𝑖𝑚𝑒𝑠 𝑛 − 𝑟 𝑡𝑖𝑚𝑒𝑠
Since the trials are independent, probability associated with this sequence is prqn-r. Exactly the
same probability can be associated with any sequence for which X = r. Number of such sequences
is nCr since we can choose r successes from n trials in nC
r ways. Hence the theorem.

Definition: A discrete random variable is said to follow Binomial distribution if its pmf is given
by
𝑛
𝐶𝑟 𝑝𝑟 𝑞 𝑛−𝑟 𝑓𝑜𝑟 𝑟 = 0, 1, … , 𝑛
Pr{𝑋 = 𝑟} {
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Observe that the probabilities are the coefficients in the expansion of (𝑝 + 𝑞)𝑛 and accordingly
the distribution is called the Binomial distribution.
Note: Binomial Expansion is (𝒙 + 𝒚)𝒏 = ∑𝒏𝒓=𝟎 𝒏𝑪𝒓 𝒙𝒓 𝒚𝒏−𝒓

Mean and Variance of Binomial Distribution

Mean 𝜇 = ∑𝑥𝑖 𝑃(𝑥𝑖 ) = ∑𝑛𝑥=0 𝑥 𝑛𝐶𝑥 𝑝 𝑥 𝑞𝑛−𝑥
𝑛!
= ∑𝑛𝑥=0 𝑥 𝑥!(𝑛−𝑥)! 𝑝 𝑥 𝑞 𝑛−𝑥
𝑛!
= ∑𝑛𝑥=1 (𝑥−1)!(𝑛−𝑥)!
𝑝 𝑥 𝑞 𝑛−𝑥
(𝑛−1)!
= 𝑛 ∑𝑛𝑥=1 (𝑥−1)! 𝑝 𝑝 𝑥 𝑞 𝑛−1−(𝑥−1)
[(𝑛−1)−(𝑥−1)]!

= 𝑛𝑝 ∑𝑛𝑥=1 𝑛−1𝐶𝑟−1 𝑝 𝑥−1 𝑞 𝑛−1−(𝑥−1)

= 𝑛𝑝(𝑞 + 𝑝)𝑛−1 = 𝑛𝑝.
Variance 𝜎 2 = ∑(𝑥𝑖 − 𝜇)2 𝑃(𝑥𝑖 )
= ∑𝑛𝑥=0(𝑥 − 𝜇)2 𝑃(𝑥)
= ∑𝑛𝑥=0(𝑥 2 − 2𝑥𝜇 + 𝜇2 )𝑃(𝑥)
= ∑𝑛𝑥=0 𝑥 2 𝑃(𝑥) − 2 ∑𝑛𝑥=0 𝑥 𝜇 𝑃(𝑥) + ∑𝑛𝑥=0 𝜇2 𝑃(𝑥)
= ∑𝑛𝑥=0 𝑥 2 𝑛𝐶𝑥 𝑝 𝑥 𝑞𝑛−𝑥 − 2𝜇2 + 𝜇2
= ∑𝑛𝑥=0[𝑥(𝑥 − 1) + 𝑥] 𝑛𝐶𝑥 𝑝 𝑥 𝑞𝑛−𝑥 − 𝜇2
𝑛!
= ∑𝑛𝑥=0 𝑥(𝑥 − 1) (𝑛−𝑥)!𝑥! 𝑝 𝑥 𝑞 𝑛−𝑥 + 𝜇 − 𝜇2
(𝑛−2)!
= 𝑛(𝑛 − 1)𝑝2 ∑𝑛𝑥=2 [𝑛−2−(𝑥−2)]!(𝑥−2)! 𝑝 𝑥−2 𝑞𝑛−2−(𝑥−2) + 𝜇 − 𝜇2

= 𝑛(𝑛 − 1)𝑝2 (𝑞 + 𝑝)𝑛−2 + 𝜇 − 𝜇2

= 𝑛(𝑛 − 1)𝑝2 + 𝜇 − 𝜇2
= 𝑛2 𝑝2 − 𝑛𝑝2 + 𝑛𝑝 − 𝑛2 𝑝2 = 𝑛𝑝(1 − 𝑝) = 𝑛𝑝𝑞.
Standard deviation 𝜎 = √𝑛𝑝𝑞.

2
Examples:
1
1. The probability that a pen manufactured by a company will be defective is . If 12 such pens
10

are manufactured, find the probability that

a) exactly 2 b) at least 2 c) none
will be defective
Solution:
Let X = number of defective pens. The probability of a defective pen = 0.1
Then XB(12, 0.1).
a) Probability that exactly 2 will be defective = Pr{X=2} = 12𝐶 2 𝑝2 𝑞𝑛−2
= 12𝐶2 𝑝2 𝑞12−2
= 12𝐶2 𝑝2 𝑞10 = 12𝐶2 (0.1)2 (0.9)10 = 0.2301.
b) The probability that at least 2 will be defective=1-probability that less than 2 are
defective
=1-[12𝐶 0 𝑝0 𝑞12 + 12𝐶1 𝑝1 𝑞11 ]
=1-[(0.9)12 + 12(0.1)(0.9)11 ]
=0.3412.
c) The probability that none will be defective =12𝐶0 𝑝0 𝑞12 = (0.9)12 = 0.2824.

2. 4 coins are tossed 100 times and the following results were obtained. Fit a binomial
distribution for the data and calculate the theoretical frequency (corresponding probability
frequencies).
No. of heads 0 1 2 3 4
Frequency 5 29 36 25 5

Solution: Let X be random variable denoting no. of heads. If 𝑝 is the probability of head,
then 𝑞 represents the probability of tail.
∑𝑓𝑖 𝑥𝑖 196
From the given frequency distribution, mean 𝜇 = = = 1.96.
∑𝑓𝑖 100

But 𝜇 = 𝑛𝑝 = 1.96
1.96 1.96
𝑝= = = 0.49
𝑛 4
∴ 𝑞 = 1 − 𝑝 = 1 − 0.49 = 0.51.
∴ The binomial distribution fit for this data is 100(0.49 + 0.51)4.
Hence, the corresponding probability frequencies are obtained as
100 4𝐶0 (0.49)0 (0.51)4 = 7
100 4𝐶1 (0.49)1 (0.51)3 = 26
100 4𝐶2 (0.49)2 (0.51)2 = 37

3
 100 4𝐶3 (0.49)3 (0.51)1 = 24
100 4𝐶4 (0.49)4 (0.51)0 = 6
Thus the theoretical frequencies are 7, 26, 37, 24, 6.

3. Out of 800 families with 5 children each, how many would have expect to have a) 3 boys b)
5 girls c) either 2 or 3 boys? Assume equal probability for boys and girls.
1 1
Solution: Let 𝑝 = probability of a boy=2 and 𝑞 = probability of girl= 2.

a) Probability of a family having 3 boys=5𝐶3 (0.5)3 (0.5)2 = 0.3125.

∴Expected no. of families=800 ×0.3125=250.
b) Probability of a family having 5 girls=5𝐶 0 (0.5)0 (0.5)5 = 0.03125.
∴ Expected no. of families=800×0.03125=25.
c) Probability of a family having either 2 or 3 boys= 5𝐶2 (0.5)2 (0.5)3 + 5𝐶3 (0.5)3 (0.5)2
= 0.625.
∴ Expected no. of families = 800 ×0.625=500.

4. The number of telephone lines busy at a particular time is a binomial variable with
probability 0.1 that a line is busy. If 10 lines are selected at random, what is the
probability that i) no line is busy ii) at least one line is busy iii) at most 2 lines are busy.
Solution: By data we have 𝑝 = 0.1 , 𝑞 = 1 − 𝑝 = 0.9. Also n=10.
∴ Out of 10 lines, the probability that 𝑟 lines are busy is given by 10𝐶𝑟 (0.1)𝑟 (0.9)10−𝑟 .
i) No. line is busy =10𝐶0 (0.1)0 (0.9)10 = 0.3487.
ii) Probability that at least one line is busy =1- probability that no. line is busy.
= 1 − 0.3487 = 0.6513.
iii) At most 2 lines are busy = 10𝐶0 (0.1)0 (0.9)10 + 10𝐶1 (0.1)(0.9)9 +
10𝐶2 (0.1)2 (0.9)8=0.929830.
5. In a bombing action, there is 50% chance that any bomb will strike the target. Two direct
hits are needed to destroy the target completely. How many bombs are required to be
dropped to give 99% chance or better of destroying the target?
Solution: From data, 𝑝 = 0.5 and 𝑞 = 0.5.
Where 𝑝 is the probability that bomb strike the target.
𝑃(𝑥) = 𝑛𝐶𝑥 (0.5)𝑥 (0.5)𝑛−𝑥 = 𝑛𝐶𝑥 (0.5)𝑛 represents the probability that 𝑥 bombs out of n strikes
the target.
We need to find minimum value of 𝑛 such that 2 ≤ 𝑘 ≤ 𝑛 bombs destroy the target
completely.
i.e, 𝑃(2 ≤ 𝑘 ≤ 𝑛) ≥ 0.99
∑𝑛𝐶𝑥 (0.5)𝑛 ≥ 0.99
4
 𝑛𝐶0 (0.5)𝑛 + 𝑛𝐶1 (0.5)𝑛 ≤ 1 − 0.99
1+𝑛
2𝑛
≤ 0.01

i.e., 100(1 + 𝑛) ≤ 2𝑛
till 𝑛 = 10, we get 100(1 + 𝑛) > 2𝑛
i.e., for 𝑛 = 10, 100(1 + 10) = 1100 > 210
for n=11, we get 100(1 + 𝑛) = 1200 < 211
∴ 𝑛 = 11
Minimum 11 bombs are required.

6. An airline knows that 5% of the people making reservation on a certain flight will not turn
up. Consequently their policy is to sell 52 tickets for a flight that can hold only 50
passengers. What is the probability that there will be a seat available for every passengers
who turns up.
Solution: Probability that passenger will not turn up=0.05.
i.e., = 0.05 ,𝑞 = 1 − 0.05 = 0.95.
x=no. of passenger will not turn up, 𝑛 = 52.
𝑃(𝑥) = 52𝐶𝑥 (0.05)𝑥 (0.95)52−𝑥
Required probability = 𝑃(𝑥 ≥ 50) = 1 − 𝑃(𝑥 > 50)
= 1 − {𝑃(𝑥 = 51) + 𝑃(𝑥 = 52)}
= 1 − {52𝐶51 (0.05)51 (0.95)1 + 52𝐶52 (0.05)52 (0.95)0 }.

7. If 20% of the bolts produced by a machine are defective, determine the probability that out of
4 bolts chosen at random, (a) 1 (b) 0 (c) less than 2, bolts will be defective.

Solution: The probability of a defective bolt is p = 0.2, of a non - defective bolt is

q = 1 - p = 0.8.
Let the random variable X be the number of defective bolts. Then
(a) P(X = 1) = 4C1 (0.2)1 (0.8)3  0.4096

(b) P(X = 0) = 4C0 (0.2)0 (0.8)4  0.4096

(c) P(X < 2) = P(X = 0) + P(X = 1) = 0.4096 + 0.4096 = 0.8192.

8. Let the probability that the birth weight (in grams) of babies in America is less than 2547
grams be 0.1. If X equals the number of babies that weigh less than 2547 grams at birth
among 20 of these babies selected at random, then what is P(X ≤ 3)?

5
 Solution: If a baby weighs less than 2547, then it is a success; otherwise it is a failure.
Thus X is a binomial random variable with probability of success p and n = 20. We are
3
given that p = 0.1. Hence P(X ≤ 3) = 
k 0
Ck (0.1)k (0.9)20k = 0.867
20

9. A gambler plays roulette at Monte Carlo and continues gambling, wagering the same
18
amount each time on “Red”, until he wins for the first time. If the probability of “Red” is 38

and the gambler has only enough money for 5 trials, then (a) what is the probability that
he will win before he exhausts his funds; (b) what is the probability that he wins on the
second trial?
18
Solution: p = P(Red) = 38
.

(a) Hence the probability that he will win before he exhausts his funds is given by
18
P(X ≤ 5) = 1 - P(X ≥ 6) = 1 - (1 − 𝑝)5 = 1 - (1 − 38)5 = 0.956.

(b) Similarly, the probability that he wins on the second trial is given by
18 18
P(X = 2) = (1 – p) p = (1 - )
38 38
= 0.2493.

x
 , 0 x2
10. A continuous random variable has pdf f ( x)   2 .
0, elsewhere
(a) If two independent determination of X are made, then what is the probability that
both these determinations will be greater than 1.
(b) If three independent determinations are made, what is the probability that at least 2
of these are greater than 1.
Solution:

x
 , 0 x2
f ( x)   2
0, elsewhere
Z : Number of independent determination greater than 1.
2𝑥
p = P(X > 1) = ∫1 2
𝑑𝑥 = 0.75

(a) P(Z = z) = nCz p z (1  p)n  z

9
P(Z = 2) = 16
.

(b) n = 3, P(Z ≥ 2) = P(Z = 2) + P(Z = 3)

= 3C2 (0.75)2 (0.25)32  3C3 (0.75)3 (0.25)0
27
= 32
.

6
The Poisson Distribution:
A discrete random variable X is said to follow Poisson distribution with parameter  if its pmf is
given by
𝜆𝑥 𝑒 −𝜆
𝑃(𝑥) = 𝑥!
, x = 0, 1, 2, …..

The Poisson distribution gives the number of events occurring in a given time period, given the
average number of times the event occurs over that time period.

The Poisson distribution is useful to model events such as

 Number of persons born blind per year in a city
 Number of hits websites get in a particular minute.
 Number of aircraft/road accidents in any time interval.
 Number of arrivals in any service facility like at an ATM, bank, railway station, petrol
pump, …
 Number of patients in a doctor's clinic.
 Number of meteorites greater than 1 meter diameter that strike Earth in a year
 Number of laser photons hitting a detector in a particular time interval

Mean and Variance of Poisson Distribution:

Mean 𝜇 = ∑𝑥𝑖 𝑃(𝑥𝑖 )
𝜆𝑥 𝑒 −𝜆
= ∑∞
𝑥=0 𝑥
𝑥!
𝜆𝑥
= 𝑒 −𝜆 ∑∞
𝑥=0 (𝑥−1)!

𝜆𝑥−1
= 𝑒 −𝜆 𝜆 ∑∞
𝑥=0 (𝑥−1)!

𝜆 𝜆2
= 𝑒 −𝜆 𝜆 [1 + 1! + 2!
+ ⋯ ] = 𝑒 −𝜆 𝜆𝑒 𝜆 = 𝜆.

Variance 𝜎 2 = ∑(𝑥 − 𝜇)2 𝑃(𝑥)

= ∑(𝑥 2 − 2𝜇𝑥 + 𝜇2 )𝑃(𝑥)
= ∑∞ 2 ∞
𝑥=0 𝑥 𝑃(𝑥) − 2𝜇 ∑𝑥=0 𝑥 𝑃(𝑥) + 𝜇
2

= ∑∞
𝑥=0[𝑥(𝑥 − 1) + 𝑥]𝑃(𝑥) − 𝜆
2

= ∑∞
𝑥=0 𝑥(𝑥 − 1)𝑃(𝑥) + ∑𝑥𝑃(𝑥) − 𝜆
2

𝑥(𝑥−1)𝜆𝑥 𝑒 −𝜆
= ∑∞
𝑥=0 + 𝜆 − 𝜆2
𝑥!
𝜆𝑥 𝑒 −𝜆
= ∑∞
𝑥=2 (𝑥−2)! + 𝜆 − 𝜆
2

𝜆𝑥−2
𝜎 2 = 𝜆2 ∑∞
𝑥=2 (𝑥−2)! 𝑒
−𝜆
+ 𝜆 − 𝜆2

= 𝜆2 𝑒 −𝜆 𝑒 𝜆 + 𝜆 − 𝜆2 = 𝜆
Standard deviation 𝜎 = √𝜆.

7
Theorem: Let X be a binomially distributed random variable with parameters n and p.
Pr{𝑋 = 𝑥} = nCxpxqn-x, x = 0, 1, 2, …, n
Suppose that, as n  , p  0 in such a way that np (a constant), then
− 𝑥
𝑙𝑖𝑚 𝑒 
Pr{𝑋 = 𝑥} =
𝑛→∞ 𝑥!

Proof: If X is binomial distributed, and if we put np = , then

n(n  1)(n  2). . .  n  k  1 x   

n x n x
  
x

P (X = x) = nCx p x (1  p)n  x = nCx   1   =  1  

n  n x !n x  n

 1  2   x  1 
1  1   . . . 1  
n  x    n x
= 
n  n  
 1  
x!  n
As n  

 1  2   x  1 
1  1   . . . 1   1
 n  n   n 
while

   n x    n     x 
1    1   1    e
 n  n  n
𝑢 𝑛
using the well-known result from calculus that lim (1 + ) = eu
𝑛→∞ 𝑛

𝑙𝑖𝑚 𝑒 − 𝑥
Thus, in the limiting case, we get Pr{𝑋 = 𝑥} = 𝑥! , which is the Poisson distribution.
𝑛→∞
Binomial distribution tends to Poisson distribution under the following conditions:
1) n, the number of repetitions is very large
2) p, the probability of success is very small
3) As 𝑛 → ∞, 𝑝 → 0, in such a way that np  , remains a constant.

Problems on Poisson distribution:

1. 𝑋 is a Poisson variable and it is found that the probability that 𝑋 = 2 is two third of
probability that 𝑋 = 1. Find the probability that 𝑋 = 0 and 𝑋 = 3. What is the probability
that 𝑋 exceeds 3?
𝜆𝑥 𝑒 −𝜆
Solution: 𝑃(𝑥) = 𝑥!
,𝑥 = 0,1,2, …
2
∴ 𝑃{𝑋 = 2} = 3 𝑃{𝑋 = 1}
3𝜆2 𝑒 −𝜆 2𝜆𝑒 −𝜆
i.e., =
2! 1!

3𝜆2 − 4𝜆 = 0
𝜆(3𝜆 − 4) = 0
4
𝜆 = 0 or 𝜆 = 3

8
 If 𝜆 = 0, then 𝑃(𝑥) = 0.
4 𝑥 4
4 ( ) 𝑒 −3
∴ 𝜆 ≠ 0 𝑎𝑛𝑑 𝜆 = => 𝑃(𝑥) = 3
3 𝑥!
4 0 −4 4
𝑃{𝑋 = 0} = ( ) 𝑒 3 = 𝑒 −3 = 0.26359
3
4 0 4
𝑃{𝑋 = 3} = ( ) 𝑒 −3 = 0.10413714
3
𝑃{𝑋 > 3} = 1 − 𝑃{𝑋 ≤ 3} = 1 − [𝑃{𝑋 = 0} + 𝑃{𝑋 = 1} + 𝑃{𝑋 = 2} + 𝑃{𝑋 = 3}
4 0 4 4 1 4 4 2 4 4 3 4
(3) 𝑒 −3 (3) 𝑒 −3 (3) 𝑒 −3 (3) 𝑒 −3
=1−[ + + + ]
0! 1! 2! 3!

= 1 − 0.953505 = 0.046494.
2. A manufacturer knows that the condensers that he makes contain on the average 1%
defective. He packs them in box of 100. What is the probability that a box picked at
random will contain 3 or more faulty condensers?

Solution: Let p=probability of a defective=0.01

𝑛 = 100, 𝑚𝑒𝑎𝑛 𝜆 = 𝑛𝑝 = 100 × 0.01 = 1.
∴ Probability that a box will contain 3 or more faulty condensers
= 1 − 𝑃{𝑋 < 3}
𝜆0 𝑒 −𝜆 𝜆1 𝑒 −𝜆 𝜆2 𝑒 −𝜆
=1−[ 0!
+ 1!
+ 2!
]

𝑒 −1
= 1 [𝑒 −1 + 𝑒 −1 + ]
2!
5𝑒 −1
= 1− 2
= 0.0803013

3. The number of accidents in a year to taxi drivers in a city follows a Poisson distribution
with mean 3 out of 1000 taxi drivers. Find approximately the number of the drivers with i)
no accident in a year ii) more than 3 accidents in a year.
3𝑥 𝑒 −3
Solution: 𝑃(𝑥) = 𝑥!
gives the probability of accidents to taxi drivers.

Approximate number of drivers out of 1000 with x accidents= 1000𝑃(𝑥).

i) No. of drivers with no accidents = 1000 × 𝑃(0)
30 𝑒 −3
= 1000 = 49.7870 ≅ 50.
0!
ii) No. of drivers with more than 3 accidents in a year=1000 −no. of drivers with
less than or equal to 3 accidents in a year.
= 1000 − 1000[𝑃(0) + 𝑃(1) + 𝑃(2) + 𝑃(3)]
30 𝑒 −3 3𝑒 −3 32 𝑒 −3 33 2−3
= 1000 − 1000 [ + + + ]
0! 1! 2! 3!

9
 = 353.
4. A bag contains 1 red and 7 white marbles. A marble is drawn from the bag and its color is
observed. Then the marble is put back into the bag and the contents are thoroughly
mixed. Find the probability that in 8 such drawings, a red ball is selected exactly 3 times?
Solution: Let 𝑋 be a random variable denoting the number of times red ball is selected
in 8 drawings.
1
By data, mean 𝜆 = 𝑛𝑝 = 8 × 8 = 1.
1𝑥 𝑒 −1
𝑃(𝑥) = 𝑥!
, 𝑥 = 0,1,2,3, …
13 𝑒 −1
∴Probability of selecting a red ball exactly 3 times = 𝑃{𝑋 = 3} = 3!
= 0.06131.

5. Suppose that 0.01% of the population of the city with population 10,000 suffers from
certain disease. Find the probability that there is at least two persons, who suffer from the
disease. If there are 10 such cities in state, what is the probability that at least one city
will have at least one person who suffer from the disease.
0.01
Solution: 𝑛 = 10,000, 𝑝 = 100
= 0.0001, 𝜆 = 𝑛𝑝 = 10,000 × 0.0001 = 1

∴ 𝑃(𝑋 ≥ 2) = 1 − 𝑃(𝑋 < 2)

= 1 − [𝑃(𝑋 = 0) + 𝑃(𝑋 = 1)]
𝑒 −1 𝜆0 𝑒 −1 𝜆1 2
=1−( + ) = 1 − = 0.2642.
0! 1! 𝑒
No. of cities, 𝑛 = 10.
For each city 𝑃(Person suffer from disease)=𝑃(𝑋 ≥ 1) = 1 − 𝑃(𝑋 < 1)
= 1 − 𝑃(𝑋 = 0)=0.6321.
Probability of at least one suffer from each city = 0.6321.
𝑝 = 0.6321, 𝑛 = 10, 𝑞 = 1 − 𝑝 = 1 − 0.6321 = 0.3679.
Using binomial distribution,
Probability that at least one city will have at least one person who suffer from
disease = 𝑃(𝑋 ≥ 1) = 1 − 𝑃(𝑋 = 0) = 1 − 10𝐶0 𝑝0 𝑞10 = 0.999.

6. An insurance company has discovered that only about 0.1% of the population is involved
in a certain type of accident each year. If its 10,000 policy holders were randomly selected
from the population, what is the probability that not more than 5 of its clients are
involved in such an accident next year?
Solution: Let X be a number of clients involved in accidents.
𝑛 = 10,000 , 𝑝 = 0,0001, 𝜆 = 𝑛𝑝 = 10
𝑒 −𝜆 𝜆𝑥
𝑃(𝑋 = 𝑥) = , 𝜆 = 10
𝑥!
𝑃(𝑋 ≤ 5) = 𝑒 −10 [𝑃{𝑥 = 1} + 𝑃{𝑋 = 2} + 𝑃{𝑋 = 3} + 𝑃{𝑋 = 4} + 𝑃{𝑋 = 5}
= 0.0671.

10
 7. The number of traffic accidents per week in a small city has a Poisson distribution with
mean equal to 3. What is the probability of exactly 2 accidents occur in 2 weeks?
Solution: The mean traffic accident is 3. Thus, the mean accidents in two weeks are
 = (3) (2) = 6.
 x e 
Since f ( x) 
x!
62 e6
f (2)   18 e6 .
2!

8. The distributor of been seeds determines from extensive test that 5% of large batch of seeds
won’t germinates. He sells seeds in packets of 50 and guarantees 90% germination.
Determine the probability that particular packet violate the guaranty.

Solution: p = 0.05, n = 50,  = 50 × 0.05 = 2.5.

X = Number of seeds that do not germinate
5
2.5x e2.5
P(X > 10% of 50) = P(X > 5) = 1 – P(X ≤ 5) = 1  0 x! = 0.042

Continuous Distributions
The Normal Distribution
A continuous random variable X is said to follow normal distribution with parameters 𝜇 and
𝜎 2 , if its pdf is of the form
 (x   )2
1
f ( x)  e 2 2
,    x   , −∞ < 𝜇 < ∞, 𝜎 > 0.
 2
and we denote this by 𝑋~𝑁(𝜇, 𝜎 2 ).
While statisticians and mathematicians use the term "normal distribution" for this
distribution, physicists sometimes call it a Gaussian distribution and, because of its
curved flaring shape, social scientists refer to it as the "bell curve."
Heights of people, size of items produced by machines, errors in measurements, blood
pressure, marks scored by students in a test are some examples that follow normal
distribution.
Mean of the normal distribution is 𝜇 and variance is 𝜎 2 . Also, mean, median and mode
coincide for this distribution.

11
The graph of the pdf is called the normal curve. It is bell shaped and is symmetric about the
line x= 𝜇.

x


Standard Normal Random Variable:

A normal random variable Z with mean 0 and variance 1 is called a standard normal random
variable and we denote this by Z  N(0, 1).
Its pdf is given by
1 2
𝜑(𝑧) = 𝑒 −𝑧 /2 , −∞ <𝑧< ∞
√2𝜋

The graph of 𝜑(𝑧) is called standard normal curve and it is symmetric about y – axis.

The cdf is given by (z) = Pr{ Z  z}

 (a) = Pr{Z  a}
 Pr{a  Z  b} = (b) - (a)
 (-a) = Pr{Z  -a}
= Pr{Z  a}
= 1- Pr{ Z  a}
(-a) = 1 - (a)

12
 𝑋−𝜇
If X N(, 2) , then we can show that 𝑍 = 𝜎
has N(0, 1).
𝑎−𝜇 𝑏−𝜇
Pr{a < X < b} = Pr{ <𝑍< }
𝜎 𝜎

The cdf of N(0, 1) are tabulated in the normal table and table helps in evaluating probabilities.

13
 −𝜎 𝑋−𝜇 𝜎
 Pr{- <X -  < } = Pr{ < < }= Pr{-1 < Z < 1} = (1) - (-1) = 0.6826
𝜎 𝜎 𝜎

X assumes 68.27% of its values in the interval ( - ,  + )

−2𝜎 𝑋−𝜇 2𝜎
 Pr{-2 <X -  < 2} = Pr{ < < }= Pr{-2 < Z < 2} = (2) - (-2)=0.9545
𝜎 𝜎 𝜎

X assumes 95.45% of its values in the interval ( - 2,  + 2)

−3𝜎 𝑋−𝜇 3𝜎
 Pr{-3 <X -  < 3} = Pr{ < < }= Pr{-3 < Z < 3} = (3) - (-3)=0.9973
𝜎 𝜎 𝜎

X assumes 99.7% of its values in the interval ( - 3,  + 3)

Hence values in the normal table vary only from 0.00 to 3.09

Examples 1:
P(z  1.66)  1  P(z  1.66)  1  0.9515  0.0485. (Using standard distribution table).

Examples 2:
P(1.96  z  1.96)  F (1.96)  F(1.96)  F (1.96)  [1  F (1.96)]  0.9750  (1  0.9750)  0.95
(Using standard distribution table).

Problems on Normal distribution:

1. In a normal distribution, 31% of the items are under 45 and 8% are over 64. Find the
mean and variance of distribution.
Solution: Given 𝑃{𝑋 < 45} = 31% and 𝑃{𝑋 > 64} = 8%
𝑋−𝜇 45−𝜇
i.e., 𝑃 { 𝜎
< 𝜎
} = 0.31
45−𝜇
𝑃 {𝑍 < 𝜎
} = 0.31
45−𝜇
𝜙( 𝜎
) = 0.31
−45+𝜇
1−𝑃( 𝜎
) = 0.31

14
 𝜇−45
i.e., 𝜙 ( 𝜎
) = 0.69.
𝜇−45
= 0.5
𝜎

𝜇 − 0.5𝜎 = 45 --------------------(1)
𝑃{𝑋 > 64} = 8%
𝑋−𝜇 64−𝜇
i.e., 𝑃 > { 𝜎
> 𝜎
}
64−𝜇
𝑃 {𝑍 > } = 0.08
𝜎
64−𝜇
i.e., 1 − 𝜙 ( 𝜎
) = 0.08
64−𝜇
𝜙( 𝜎
) = 0.92
64−𝜇
𝜎
= 1.4

𝜇 + 1.4𝜎 = 64-------------------(2)
From (1) and (2), we get 𝜇 = 50 and 𝜎 = 10.

2. Suppose 𝑋 has the distribution 𝑁(𝜇, 𝜎 2 ). Determine 𝑐 as function of 𝜇 and 𝜎

such that 𝑃{𝑋 ≤ 𝑐} = 2𝑃{𝑋 > 𝐶}.
Solution: 𝑃{𝑋 ≤ 𝑐} = 2𝑃{𝑋 > 𝐶}
i.e., 𝑃{𝑋 ≤ 𝑐} = 2(1 − 𝑃{𝑋 ≤ 𝑐})
𝑃{𝑋 ≤ 𝑐} = 2 − 2𝑃{𝑋 ≤ 𝑐}
3𝑃{𝑋 ≤ 𝑐} = 2
2
𝑃{𝑋 ≤ 𝑐} = 3 = 0.666
𝑋−𝜇 𝑐−𝜇
i.e., 𝑃 { 𝜎
≤ 𝜎
} = 0.666
𝑐−𝜇
𝑃 {𝑍 ≤ 𝜎
}=0.666
𝑐−𝜇
𝜙( 𝜎
) = 0.666
𝑐−𝜇
𝜎
= 0.43

𝑐 = 𝜇 + 0.43.
3. Suppose that life length of two electronic device say 𝐷1 and 𝐷2 have distributions 𝑁(40,36)
and 𝑁(45,9) respectively. If the electronic device is to be used for 45 hours period, which
device is to be preferred? If it is to be used for 48 hours period, which device is to be
preferred?
Solution: Let 𝑋 represents the life lengths of the electronic device.
For device 𝐷1 ,
(i) For a period of 45 hours,
𝑃{𝑋 ≥ 45} = 1 − 𝑃{𝑋 < 45} = 1 − 𝑃{𝑍 − 0.8333} = 1 − 𝜙(0.8333)
= 1 − 0.7967 = 0.2053 .
(ii) For a period of 48 hours,

15
 𝑃{𝑋 ≥ 48} = 1 − 𝑃{𝑋 < 48} = 1 − 𝑃{𝑍 < 1.333} = 1 − 𝜙(1.333)
= 1 − 0.9082 = 0.0918.
For device 𝐷2 ,
(i) For a period of 45 hours,
𝑃{𝑋 ≥ 45} = 1 − 𝑃{𝑋 < 45} = 1 − 𝑃{𝑍 < 0} = 1 − 𝜙(0) = 1 − 0.5
= 0.5.
(ii) For a period of 48 hours,
𝑃{𝑋 ≥ 48} = 1 − 𝑃{𝑋 < 48} = 1 − 𝑃{𝑍 < 1} = 1 − 𝜙(1)
= 1 − 0.8413 = 0.1587.
For a period of 45 hours, device 𝐷2 is to be preferred whereas for a period of 48 hours,
device 𝐷2 is preferred.
4. For a normally distributed variate with mean 1 and standard deviation 3. Find the
probabilities that i) 3.43 ≤ 𝑋 ≤ 6.19 ii) −1.43 ≤ 𝑋 ≤ 6.19
Solution:
3.43−1 6.19−1
(i) 𝑃{3.43 ≤ 𝑋 ≤ 6.19} = 𝑃 { 3
≤𝑍≤ 3
}

= 𝑃{0.81 ≤ 𝑍 ≤ 1.73}
= 𝜙(1.73) − 𝜙(0.81)
= 0.9582 − 0.7910 = 0.1672.
−1.43−1 6.19−1
(ii)𝑃{−1.43 ≤ 𝑋 ≤ 6.19} = 𝑃 { 3
≤𝑍≤ 3
}

= 𝑃{−0.81 ≤ 𝑍 ≤ 1.73}
= 𝜙(1.73) − 𝜙(−0.81)
= 𝜙(1.73) − (1 − 𝜙(0.81))
= 0.9582 + 0.7910 − 1 = 0.7492.

5. A fair coin is tossed 500 times. Find the probabilities that the number of heads will not
differ from 250 by (a) more than 10 (b) more than 30.
Solution: Let 𝑋 represents the number of heads in 500 tosses.
1
∴ 𝑝 =probability of head turning up= 2 and 𝑛 = 500.

1 1 1
Mean= 𝑛𝑝 = 500 × 2 = 250 and 𝜎 = √𝑛𝑝𝑞 = √500 × 2 × 2 = 11.18033.

Since 𝑛 is very large and 𝑝 is not small. We can use normal approximation to
binomial.
a) Probability that number of heads will not differ from 250 by more than 1=
240−250 260−250
𝑃{240 ≤ 𝑋 ≤ 260} = 𝑃 {11.18033 ≤ 𝑍 ≤ 11.18033}

= 𝑃{−0.894427191 ≤ 𝑍 ≤ 0.894427191}
= 2𝜙(0.894427191) − 1

16
 = 2(0.8133) − 1 = 0.6266 .
b) Probability that number of heads will not differ from 250 by more than 30=
220 − 250 280 − 250
𝑃{220 ≤ 𝑋 ≤ 280} = 𝑃 { ≤𝑍≤ }
11.180339 11.180339
= 𝑃{−2.68328172 ≤ 𝑍 ≤ 2.68328172}
= 2𝜙(2.68328172) − 1
= 2 × 0.9963 − 1 = 0.9920.

6. The weekly wages of workers in a certain factory was found to be normally distributed
with mean 𝑅𝑠. 500 and standard deviation 𝑅𝑠. 50. There are 228 persons getting at least
𝑅𝑠. 600. Find the number of workers in the factory.
600−500
Solution: For 𝑃{𝑋 ≤ 600} = 𝑃 {𝑍 ≤ } = 𝑃{𝑍 ≤ 2} = 1 − 𝑃{𝑍 ≤ 2}
50

𝑃{𝑋 ≥ 600} = 1 − 𝜙(𝑍) = 0.0228.

If 𝑛 is total number of workers then 𝑛𝑃{𝑋 ≥ 600} = 228.
i.e., 𝑛 × 0.0228 = 228.
228
𝑛 = 0.0228 = 10,000.

There are 10,000 workers in the factory.

7. If X is normally distributed with N(1, 4). Find P(|X| > 4).

x
Solution: If z  , then,


P(| X |  4)  1  P(| X |  4)  1  P( 4  X  4)
 4  1 4 1 
 1 P  Z   1   F (1.5)  F (2.5) 
 2 2 
 2  0.9332  0.9938  0.073

8. In a normal distribution 31% of items rare under 45 and 48% are over 64. Find mean and
standard deviation.
Solution:
X = 45, area = 31%
 Z  0.5
X  45  
Z   0.5 
 
 0.5  45   ... . (1)

X = 64 , area = 48%

17
  Z  0.05
X  64  
Z  0.05 
 
 0.05   64   ... . (2)

Solving (1) and (2),

  62.27,   34.54.

9. If mean marks is 60 and standard deviation is 10, 70% failed in examination. What is the
grace marks given to obtain 70% pass the examination?

Solution:
  60,   10.
P( Z  a )  0.7  Z  0.525
X  X  60
Z  0.525 
 10
 X  65.25

P( Z  a )  0.3  Z  0.525
X  X  60
Z   0.525 
 10
 X  54.75

Therefore, grace mark = 65.25 – 54.75 = 10.5

10. Local authorities in a certain city install 10000 electric lamps in streets of city. If
these lamps have average life of 1000 burning hours with standard deviation of 200 hours.
What is the number of lamps might be expected to fail (i) in first 800 hours (ii) between
800 and 1200 hours.

Solution:   1000,   200.

(a)

 800  1000 
P( Z  800)  P  Z  
 200 
 P(Z  1)
 1  P(Z  1)
 0.1587

Number of lamps = 0.1587 X 10000 = 1587.

18
 (b)

 800  1000 1200  1000 

P(800  Z  1200)  P  Z 
 200 200 
 P(1  Z  1)
 0.6826

Number of lamps = 0.6826 X 10000 = 6826

11. Annual rainfall at a place is known to be normal distributed with   29.5 inches

and   2.5 inches. How many inches of rain is expected to exceeded about 5% of the
time?
Solution:

  29.5,   2.5.
P(Z  a)  0.05
 P(Z  a)  0.95
 Z  1.645
X     Z  33.61

Therefore, amount of rain exceeded = 33.61 - 29.5 = 4.11 inches.

Exponential Distribution:
A continuous random variable X is said to have exponential distribution with parameter 𝛼 > 0 if
its pdf is given by
𝛼𝑒 −𝛼𝑥 , 𝑥 > 0
𝑓(𝑥) = {
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
and we write 𝑋~exp(𝛼)
Mean:
∞
𝐸(𝑋) = ∫0 𝑥𝛼𝑒 −𝛼𝑥 𝑑𝑥
Put 𝛼𝑥 = 𝑧
Then
1 ∞ 1 ∞ 1 1
𝐸(𝑋) = 𝛼 ∫0 𝑧𝑒 −𝑧 𝑑𝑧 = 𝛼 ∫0 𝑒 −𝑧 𝑧 2−1 𝑑𝑧 = 𝛼 Γ(2) = 𝛼

Variance:
𝑉(𝑋) = 𝐸(𝑋 2 ) − [𝐸(𝑋)]2
∞
𝐸(𝑋 2 ) = ∫0 𝑥 2 𝛼𝑒 −𝛼𝑥 𝑑𝑥
Put 𝛼𝑥 = 𝑧
1 ∞ 1 ∞ 1 2! 2
𝐸(𝑋 2 ) = 𝛼2 ∫0 𝑧 2 𝑒 −𝑧 𝑑𝑧 = 𝛼2 ∫0 𝑒 −𝑧 𝑧 3−1 𝑑𝑧 = 𝛼 Γ(3) = 𝛼2 = 𝛼2
19
 1
𝑉(𝑋) = 𝛼2

Example: The life length of a certain type of a battery is a random variable having exponential
distribution with mean 200 hours. Find the probability that such a battery will work for
(i) at most 100 hours (ii) anywhere between 400 and 600 hours
Solution:
E(X) = 200
1/ = 200   = 0.005
0.005𝑒 −0.005𝑥 , 𝑥 > 0
Hence the pdf is f(x) == {
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
100
(i) Pr{battery will work for at most 100 hours} = Pr{ X  100} =∫0 0.005𝑒 −0.005𝑥 𝑑𝑥= 0.39
600
(ii) Pr{ Battery will between 400 and 600 hours} =Pr{400  X  600} = ∫400 0.005𝑒 −0.005𝑥 𝑑𝑥= 0.0855

Gamma Distribution
A continuous random variable X is said to have Gamma distribution with parameters
𝑟 > 0 𝑎𝑛𝑑 𝛼 > 0 if its pdf is given by

𝛼 −𝛼𝑥
𝑒 (𝛼𝑥)𝑟−1 , 𝑥 > 0
𝑓(𝑥) = {Γ(𝑟)
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒

and we denote this by 𝑋~G(𝑟, 𝛼)

Mean:
∞ 𝛼
𝐸(𝑋) = ∫0 𝑥 𝑒 −𝛼𝑥 (𝛼𝑥)𝑟−1 𝑑𝑥, Put 𝛼𝑥 = 𝑧
Γ(𝑟)

Then,
1 1 ∞ 1 1 1 1 𝑟
𝐸(𝑋) = ∫ 𝑒 −𝑧 𝑧 𝑟 𝑑𝑧
𝛼 Γ(𝑟) 0
=
𝛼 Γ(𝑟)
Γ(𝑟 + 1) =
𝛼 Γ(𝑟)
rΓ(𝑟) =
𝛼

Variance:
𝑉(𝑋) = 𝐸(𝑋 2 ) − [𝐸(𝑋)]2
∞ 𝛼 𝑟 2 +𝑟
𝐸(𝑋 2 ) = ∫0 𝑥 2 Γ(𝑟) 𝑒 −𝛼𝑥 (𝛼𝑥)𝑟−1 𝑑𝑥 = 𝛼2
𝑟
𝑉(𝑋) = 𝛼2

Chi-square distribution
A continuous random variable X is said to have Chi-square distribution if its pdf is given by

1 𝑥 𝑛
𝑛 𝑒 −2 𝑥 2 −1 , 𝑥 > 0, 𝑛>0
𝑓(𝑥) = {2𝑛/2 Γ ( )
2
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
We write 𝑋~2 (𝑛), where n is called the number of degrees of freedom.

20
 𝑥 𝑛
∞ 1
Mean: 𝐸(𝑋) = ∫0 𝑥 𝑛 𝑒 −2 𝑥 2 −1 𝑑𝑥
2𝑛/2 Γ( )
2
𝑥
Put 2
=𝑧

1 ∞ 2 𝑛
Then 𝐸(𝑋) = 𝑛 ∫0 𝑒 −𝑧 (2𝑧)𝑛/2 2𝑑𝑧 = 𝑛 Γ (2 + 1) = 𝑛
2𝑛/2 Γ( ) Γ( )
2 2

Variance:
𝑉(𝑋) = 𝐸(𝑋 2 ) − [𝐸(𝑋)]2
𝑥 𝑛
∞ 1
𝐸(𝑋 2 ) = ∫0 𝑥 2 𝑛 𝑒 −2 𝑥 2 −1 𝑑𝑥 = 𝑛2 + 2𝑛
2𝑛/2 Γ( )
2

𝑉(𝑋) = 2𝑛

Note: Exponential distribution is a special case of Gamma distribution with r =1.

Chi square distribution is a special case of Gamma distribution with r = n/2 and  = ½.

21