3

Discrete Probability Distributions

Probabtlity Distributlon, BernoulIi's Distribution, Binomial Distribution.
Poisson's Distribution

$3.1. Introduction:
Earlicr we have alrcady discussed the frequcncy distributions. Here we shall
discuss the discrete probability distribution in which the variables are distributcd according
10 some definite probability law. Here we shall discuss the probability distribution functions,.
their propertics and characteristics which arc important Irom practical utility view point.

S 3.2. Probability Distribution :

Let us consider a simple example of throwing two unbiased coins simultaneously.
Let us consider the numbcr of heads () oblaincd. What are the possible values of X?
Obviously X takes the valucs 0, 1, 2.
Now
PX= 0) =P (No head) P (lail on Ist coin) P (tail on 2nd coin)
-

PW= 1) =P (l head)
=P (head on lst and tail on 2nd) + P ((ail on lst and head on 2nd)

2222 4
P(X= 2) = P (2 hcads)
-P (head on both Ist and 2nd)
Thereforc we have
2

P()
This is called probability distribution of the random variable X.
"Probability distribution is a distribution of random variable in which probabilities
e distributed to the d1fferent values ofX such that their stum is unity (1)".
58 | Statistics and Probability Theory
§ 3.3. Probability Distribution Function (PDF) :
If we coald denote the above distribution with a function of random vana.
(0c. by p (x), then p (r) is called as a probability distribution function. In other w
Mord
function p (x) of random variable x is said to be a probability distribution function
satisies two conditions

() p)20, VreR
depending upon the range of random variable x. ie. all the valucs of plx) should be grey
than or cqual to zero and sum of all the probabilitics must be cqual to one in the given ra
of valucs of x
p() is said to be a discrete PDF ifx assumes only countable valucs. But if x
rf(x) is said to be a continuo
all possible values within the given range of x, then p () orf
probability distribution function. (or Probability Density Function).

83.4. Bernoulli Distribution :

The distribution in which the random variable X has only two values 0 and
with probabilities q = -p and p is called the Bernoulli's distribution.
Mahematically, the probability distribution is
1
P () p. where ptq=l
wherep =probability of succcss, q probability of failure and p +q = 1,is called Bemoullii
It is also known as single trial distribution.
distribution ents of the rnoulli distribution: The first four moments about the ong
of the Bemoulli distriution are oblaincd as follows
H=E(X) p, x, =4.0+ p.l=p
B=Er?)-Lp, =q.0+ p.t' =p
=ELr)- Ep, -q 0+p lI'=p
Wi E(r)- Ep. =q0+p i' =p
Mun ad Variance of the Bernoulli distribution :
Mean = F== E(N)=p
Variance =o' =#; - =p- p = p(l- p)= pq
Standard deviation =G =pq
Moment Gengrating Function of Bernouli D'stribution :
MGF about the origim is defincd as
4(9) =E(e)= Ep, =qe" +pel!
=q+pe'
But
 Discrete Probability Distributions |59
$3.5. Discrete Uniform Distribution :
The distribution in which the random variable X takes the vatucs 0, I, 2,
-,n cach with the same probability, 1/n is called the discrete uniform distribution.
Mathematically.

P(X) =n+1 X =0, 1,2,....,n

Also,

Such a distribution ariscs in the tossing of a coin or casting of a dicc.

Mean and Variance
n(n+1)
Mean =F E(N)=Mi
=o+1t...*n)2(n+1)i
Now, Ë =o+1² +2' +... a2) -"o+)(2n +1)n(2n +1) 6
6(n+1)

"+) n n(n+2)
variance =#, = g' = us -u? 6 4 12
Moment Generating Function :

Mo () = E( )- ie

+....+e (n+1) 1-e

Ex. IfX has the discrete uniform distribution
P(X = k) =- for k =0,1,.
Find its MGF and hence fnd men and variance of X.
Sol. Xis uniform variate with p.m.f
P(X =k) =! for k= 0,1.
. Momcnt gencrating function of any distribution is givcn by

Mr ()= Ee".P(X =k). Here P =a constant

so for the unifom dianbuion My (0)-i i.]

Mean =
 60 | Statistics and Probability Theory

1+2+3+.n _n(n+1)

--).
S 3.6. Binomial Distribution :
The Binomial distribution is derived from
Binomial coeflicicnts in the cxpan
of (x+". It is also called as Bernoulli distribution associated with the nansc of Joo
BemoulliA Bermoulli process is a random process.
Here the process is performcd under the sanc conditions for a fixcd and fn
number of independent trailsn Let x be the number of successes
p The probability of success, qThe probability of failure = -p.
The number of successes in trials may be 0, 1, 2,.. ,r, n and is obvio
chance variate.
The probability of x successes and conscqucntly (n - x) failures in n independo
trials, in some specificd order (say) SFFSSE... SFSS is given by the compound probabi
thcorcm as
P(SFFSSF..SESS) = P(S) P () P() P5) P(S) P()x. * P (S) P() P(S
(S)
=pq4P p4.p.q Pp.
PP.p. =p' q
r times ( t

But x successes inn trials can occur in "C. vays and the probability of cach1
hese ways is p'q,
Hence the probability of x successes in n trials takcn in any order is givcu by à
ddition theorem of probability as :
B(n, p.r) =P(X = x)= P. = "C, p'y. I=0,1,2,..,n
=0, otherwise,
To verify whether B (n,p,r) denotes a probability density function, let us des1g
ead as a success and ilurc with correspouding probability p and q respcln
Jacob Bernoullie (1654-1705) was a Swiss Mathematician.
 Discrete Probability Distributions |61
Let the process be performed n tines, then the probability of successes be distributed as
follows
No. of trial_ n
No. of successes
Probability
P(X= 0) 4.q...q=q"
PX= 1) "C p.q.q=" C pg l
P(X= 2) "C; p°g
:
P(X-r)

P(X=n)
Adding all the probabilities, we get

Hence the function denoted by B (n, p: r) is a probability distribution function.

.&3.7. Mean of the Binomial probability distribution :
The mean of binomially distributed random variable X, designated by or
u, or E () is thcoretical expected number ofsuccesses in n trials i.e.

i =E() = x B(n, px) =X xP(X =x)

i.e. the mcan of Xis the sum of the products of the valucs that X can take multiplicd
by their respective probabilities.

n(n-1)!
x-0

(n-1)!
(-1)(-x) P pq
= np )
Thus the mean of binomial distribution is np.

SJ.8. Variance of the Binomial probability distribution:

The variance of binomially distributed random
vaiable X is given by
o'= Var (X)= -(4 ..(1)
62 Statistics and Probability Theory

Now ; =E(x²) -) B{n, p. x) = 2 r'P(X =*)

r=0
rx(*)+x, we have
As we can write

r(*-1) "C, p'a^+ r"C, p'g"

Procccding similarly, we get
H =n(n-1)p² +np
Thereforç a' =n(n-)p² +np-(np)
=np(1-p) =npq
distribution is G= Jnpq
Thus thc stgndard deviation of the binomial
and pBe variance o = npq)
S3.9. First Four Moments About Originmoments
:
:
We have already obtaincd first two
3rd moment about origin :

- {r(*-1)(x - 2) +3r(r- 1) +x} "C, p* q*

=n(n-1)(n - 2)p' *-C pax

=n(n-1)(n-2) p° (9+ p)+3n(n -1)p² +np

ý =n(n- )(n- 2)p + 3n(n -1) p+ np
4th Moment about origin :

2 (r(*-1)(x - 2)(x - 3) +6x(x- 1)(x- 2)

+7x(*-1) +x} "C,
= n(n-1)(n -2)(n-3)p*+6n(n-)(0-2) p
+7n(n -1)p² + np [on sinmpl
 Discrete Probability Distributions 63
First Four Moments abuut Mean : The first four moments about the mcan of the
Binomial distribution are obtaincd as follows:
H=2p, (*, -+) =Ep,, -i-F-i =0

H3 = - 3uu + 2u = npy(g-p)
H =-4ui +6uu -3unpg[1 +3(n -2) pg]
§ 3.10. Karl Pearson's coefficients of Skewness and Kurtosis for Binomial
Distribution:
n'p°g? (g-p(9-p (0-2p
npq npg

B. H4npqll+ 3(7-2)Pa]_I+3(n -2) pq.. 1-6pg

n'pq? npq npg

Y2 =B, -3 =-6pg
Vnpg npq
The binomial distribution is positively skewed ifp
ncgatively skewed ifp > 0.5 and symmctric whenp = 0.5.
(1) If Y2 >0 or l- 6pq > 0 e. pg < then the curve is leptokurtic
(2) If Y =0 or pq then the curve is mesokurtic
(3) If y2 <0 or pq> then the cuve is platykurtic.

§ 3.11. Moment Generating Function of Binomial Distribution :

By definition, MGF about origin is given by
M(1) =E()- EP(*)e - "C, p*q"e
-"C, (pe)g=(gt pe)
Calculation of Mean and Variance with the help of MGFof Binomial Distribution:
Mcan =I=u

= np(q + p) =np

=nplle+p)l +(n-1)p(g+ py
64 | Satistics and Probability Theory
-np{l+ (n-1)p)= np(np +q)
Variance =g =, = -un'p+npq -np' npg
63.12. Distcjbution Function of a Binomial Distribution : is defincd as
The distribution function F(r) of aBinomial distribution
F(r) =P(Xsr)
3.13. Probability Distribution ofa Binomial Distribution : Probablity distr
of a Binomial distribution is defincd as

PX) "c, p'g p"

It is also known as Binomial probability distribution
8 3.14. Recurrence Relation for Binomial Distribhution :

B(n, p.x +l) =p(r+l) = "C,. p--l

(m)

(r+)((a-r)

g (r+)

Hence P(r+) p)
63.15. Fitting a Binomial Distribution : obserne data, the following procode
When a distribution is to be fittcd to
adoptcd When p and g arc cqual the distribution is symme
1. Determine he valuc
for p and q may be n olpdyerinethe yaluc of any tcrm and conscq=
ends of the scrics are cqual.
terms eq the distrbution is
(b) If p< /2, the distribution is positively skewed.
skeved,
(c) If p> 12 the distribution is negatively
the
2. Expand the binomial (q+ p)". The power n is cqual to onc less than
will bes
of terms in the expanded binomial. Thus when six coins are tossed there
terms.3. Multiply cach term of the expanded binonial by N (he total frequency). in
takes the
obBain the expected frequency in cach category. The rocurrence fomula

where f(r) is the cxpected frequency,.

 | 65
Discrete Probability Distributions
(2)
For x0, 1, 2, 3.
obtaincd by putting x
From (1) and (2), all the frequencies can be
Distribution :
8 3.16. Importance of the Binomial
discrcte probability distribution
The binomial probability distribution is a life events. For example. nple. in a
which is useful in describing an cnormous variety of rcal
factory thc quality control inspector wants to know thec probability of obtaining of defcctictive

random sample of 400 bolts if I5% of thc bolts are defcctive He can quick ickb
bolts in a can be uscd whc . h.o.
It
obtain he answer from tables of the binomial probability distribution. as onc of tl
(0) The outcome of cach trial in the process and characterised
types of possible outcomes. is indcpcndent of the
(2) The possibility of outcome of any trial does change and
results of the previous trials
lustrative Examples
seven
Ex. 1. Ten coins are tossed together. Find the probability of getting atleast
heads.
Sol. The probability of getting a hcad (p) =
The probability of not getting a head (9) .
Thercfore the probability of getting at lcast seven hcads in a throw of 10 coins is
P()+ P(8) +P(9) +P(I0)

120+ 45+ 10+| 176 11

Ans.
1024 024

Ex.2. If the probabitity of defective bolts is 0.I, find the mean and standard
eviation for the distribution of defective bolts in a total of 500.
Sol. Herep= 0.1;q=0.9 and n 500
Therefore mean np = 500 x 0.1 50 Ans.
Thus we can expect 50 bolts to be defcctive.
S.D, =/4pq = 1)x(,9)xS00 =45 3W5 Ans.

Ex.3. The incidence of occupational disease in an industry is such that the

vorkers have a 20 %chance of suffereing from it. What is the probability that out of 6
workers 4 or more will suffer from disease |Raj BE, (I Sem. EE), 44 |
Sol. The probability of a man sullering from disease i.e.
20
66 Statistics and Probability Theory
The probability of a worker not sulfering from discase ie.
1 4

Thercfore the probability of 4 or more workers suffering trom disease i

P(X 24) =P()+P(5)+ P(6) =C,p'g²+ C, p'q+ Ce p expe

240 241 265

1S62515625 15625 1562s 12

Ex.4, 10 % produced by a machine are defective. Find the probal

following when they are checked at randum by examining samples of 5 :
(a) None i defective (b) one is defective (c) atmost one is defective
Sol. If getting a defective bolt is a success, then
10 q=1-0.=09
100 =0.1 and
(a) Probability of not getting any defcctive bolt in a sample of 5 bolts.
-C, a'p=(9 =(0.9=0.5905
(b) Probability of getting 1defective bolt in a sample of 5 bolts
='C g'p= Sx(0.9)* x(0.1) =0.3281
Probability of getting maximum I defcctive bolt
P(Xsi) =P(0)+P() ='Ca'p+ 'Ca'p =0 5905 +03
= 0.9186.
100 invesiy wil g
EL. SAssumining that halfof thepopulation is vegetarian and vegetarian
they are ;imane
ccn investigate the samples of 10 individuals to see whether
ure vegetarian?
many investigators will report that 3 or less people
the ta

Sol. Here we have 100 (= M) scts of 10 (= n) trials cach, the probabie

vegetarian being (p) so that q
The number of investigalors reporting 0, I, 2,...10 consumers arc given br...
successive terns of the binonial at les

first fourt
The number rcporting thrce or less vegetarians shall be the sum of
=P() +P() +P(2)+ P(3)
 Discrete Probability Distributions | 67

100 100x 176 275

J6 17 nearly Ans.
1024

6. Out of 800 familles with children each, how many families would be
expected to have (a) 2 boys and 2 girls. (b) ut leust one boy (c) at ost 2 girls.
Assume equal probabilities for boys and girls.
Sol. If a family having a boy cess, then
P= and =
(a) Probability that a family having 2 boys and 2 girls is

No. of families having 2 boys and 2 girls are

N=800-300 Ans.

(6) Probability that afamily having atleast one boy is =|-P(0)

Ans.
No. of families having atlcast one boy are = N.=750
(c) Probability that a family having at most 2 girls is
= P(2) +P(3) +P(4)
-1-{P(0)+ P(0) - 1-{, pg' +'Gi pa')
A
-1-(9+(}--#
No. of families having at most 2girls are = N = S50 Ans.

Ef7, In a precision bombing attack there is a 50% chance that any one bomb
h will strike the target Tvo direct hits are required to destroy the target completely. How
many bombs must be dropped to give a 99% chance or better for completely destroying
the target.
Sol, Given that
p= prob. that the bomb strikes the target = 50% =.
Let n be the number of bombs which should be dropped to ensure 999% chance or
better of completely destroying the target. This means the probability that out of n bombs,
at least two strike the target, is greater than 0.99.
Let X be a rando1n variate denoting the number of bombs striking the target.
Then X- B(n.x,p =}) with
x=0,1, 2..., n

According to the given problem, wvc must have P(X 22) 2 0.99
[-P(Xs1)]20.99 |-{P(0)+ P()}] 2099
 68 | Statistics and Probability Theory
l+r
20.99 0.01 2
2" x(0.01) 2l+n 2 2100+100
Clearly the minimum value ofn satisfying the inequality (1) is 11.
Therefore n=l1.
Ex. 8. Six dice are thrown 729 times. How many times do you expect at l.
dice to show a 5 or 6 ?
Sol. Herc N= 729, n 6
p be the prohal
Let the occurrence of 5 or 6 be regarded as success and
succesS.

Now p Prob. of occurrence of 5 or 6= =

q=1-p
.. Prob. of x successes is given by
66-I
P(X = x) = C, p gox= 36
By theorem of total probability, probability of at least three successes
P(X 2 3)=P(3) +P(4) +P(S) +P(6)

L[6548,65s44
- 532.1 4.3.2.1
+6.2 +1}
160 +60 + 12 + I}
. Number of times at least threc successcs occur
x729 =233.
men and 'B' women, show the
Ex. 9. If mthings are distributed amongby 'a'men is odd, is
probability that umber of things received

a(b+a)" -(b-a)"
(b+a)
Sok It is given that total number of things m which are to be distributed
r of men =a and number of women = b.

Let p be the probability that a thing is reccived by man so p = a+b

Then thc probability that a thing is rcceived by womcn.

b
a+b a+b
 Discrete Probability Distributions | 69
The probability that out of m things cxactly x are reccived by men and the rest by
women can be represented by the use of binomial distribution. It is given by
PX = x)= "C, p* qM. x= 0,1, 2,..Jm,
where p and q are given by the expressions given above.
The probability that the number of things reccived by men is odd is given by
e-p()+p(3)+p(s)+. "C qp+ "C qmp+.
But by the binomial theorem, we have

Therefore
e-lo*p-a-]
b-a
Also qtp=l, 9-p= h+a

Thereforce Hence proved.

Ex. 10. A department in a works has 10

machines which mnay need adjustment
from time to time during the day. Three of these machines are old, each having
probability of 1/1l of needing adjustnent during the day, and 7 are new having
corresponding probabilities df /21.
Assuming that no machine needs adjustment bvice on the same day, determine
the probabilities that on a particular day
(a) just 2 old and no new nmachine need adjustment.
(b) ifjust 2 machine need adjustment, they are of same type
Sol. (a) Let p, = Probability that an old machines nccds adjustment
Therefore
and p, = Probability that a new machines necds adjustment
92 =1-P -
Then P() = Probability that ' old machines nced
adjustment

... (1)
and P (r) = Probability thatr'new machines necd adjustment
(2)
(a) Putting r=2 in (l) and r= 0 in (2)
i.e. just 2 old and no new machine
Required probability = P(2). P(0)
70] Statistics and Probability Theory
wc gct 0.016
b) Putting r=0 in (1) and r= 2 in (2), we get
P(0) P,(2) =0.028
The probability that "f pust pao machines need adiustment, they are ot. 1
type is equal to the sum of two probabilitis given by (3) and (4), we get
Required probability 0,016+0,028 - 0.044

Ex. 11. Fu a binomial distribution to the following data

Variate (x) 2
Frequency (f) 10 20 25 20 17 Tu
Sol. Herc n = 5, N =Ef =100
F 1 0 x 0 + 2 0 x|+ 25x 2+ 20x3 +17x4+ 8xs
10+20+ 25 + 20+17+8
20+ 50 + 60+ 68+40
0-238
100 100
Tnp 2,38 = 5xp ’p=0.476
so that 9=1-p=|-0.476 =0.524
Nonc N(q+p)" 100(0.524 + 0.476) 3.

or S, =l00xc, (0. 524 (0 476)'

r= 0,1,2,3,4, 5
4, I8, 33, 30,), 113, 2

Ex. 12. A perfect t cubie die is thrown a large number of times in sets oft
occurrence of 5 or 6 is called a success. ln what pruportion of the sets would youc
3successes
Sol. Here n=8,
p= probability of success = probability of occurrence of 5 or61

q=l-p=l-}
: Probability of x successes is given by

:. P(r=3)- 87.6 32 1792

. Required proportion
sixs1
0273| = 27.31%
656|
0.2731
 | 71
Discrete Probubility Distributions

Exercise IiI (a)

is 2. Find the distribution
For aBinomial distribution. the mean is and variance
four or morc hcads.
2. A coin is tosscd six timcs. Find thc probabil1ls of oblaining
is 3/5. In scrics of 6 yamcs what
3: Aand Bplay a game in which A's clhances of winning
is the chance that A will win atlcast 4 gancs?
probability that thcv obtain the
Each of two persous tosscs three fair cois Whal is the
same number of cads
two times thc
Tivo dice arc throWn toycther 4 times Fnd the probabilits that atlçast
suni of lhe numbcrs is 7,
6. Eight coins arce tosscd at a time. 256 timcs Nmber of hcads observcd at cach throw
arc recorded and the results are as given bclow. Fund thc cspccted frcqucncey and fit a
Binomial distribution
What are the thcorctical values of the mcan and standard devation
No. of heads (x) 0 1 2 4 7 8 Total
No. of timcs () 2 6 30 67 56 32 | 256
7. A coin is tosscd 4 times, What is the probability of getting
(a) two hcads (b) atlcast Iwo hcads
8. The probability that a bomb dropped from a planc strike the targct is l/5. If six bombs
are droped, find the probability that atleast lwo will strike the targct and exactly two
will strike the targci
9. If the probability of a defcctive bolt is 0.1.
Find (a) the mean, and
(b) thc standard deviation of thc defçctive bol1s in a lotal of 900
c 10. If 10% of the bolts produced by a machine are defcctive, use binonial distribution to
detemine the probability that out of 5 bolts sclccted at random, atmost one will be
defective.
11. The probability that a 60 years old man will remain alive at the age of 70 is ).65. What
is the probability that atleast 7out of such 10 persons will remain alive at thc age of 70.
12. If 10 % of the pens manufactured by a company are defcclive, find the probability that
a box of 12 pens contain
(a) cxactly 2 defcctive pens (b) atleast two defective pens
13. The probability that at any moment onc telcphonc linc out of 10 will be busy is 0.2
(a) What is the probability that 5 lines are busy ?
(b) Find he expectcd number of the busy lines and also find out thc
probability of this
number,
(c) what is the probability that all the lncs arc busv?
72 ] Statistics and Probability Theory

PoissoN'S DISTRIBUTION
S 3.17. Poisson Distribution :
This is a very widely used distribution discovered by Poisson
Poisson distribution differs from the binomal dirtribution in two import:
ortat
(a) Rather than consisting of discrete trials, the distribution operates ce
over some given amount of time, distance, area etc.
(b) Rather than producing scquence of successes and failures, the
dit
produces successcs, which occur at random points in thc specificd time, distane
These successes are commonly referrcd to as 'occurrances
The Poisson distribution may be used to approximate binomal distribution
is large and p is small is regarded as the limit of binom. dictribution.
§ 3.18. Deduction of Poisson distribution as a limiting case of B
distribution.
The two postulates
ulates for the randam variablc X to be Poisson distributed are
(a) No. trials infinite ie
() Probability of success is very low i.c. p’0
(c) np = m (or A) is finite real positive number, called the parameter of
distribution.
APoisson distribution may be expectcd in the cases where the chance of any ind
event a success is small. The distribution is uscd to describe the behaviour of rare:
and has been called "the law of improbable events"
We know that the probability of x successes out of n trials is given by
B(n.p;x) = "C, pg"*

n(n-1)(1-2).(n-x+)pq
x!

Putting p=" and q=l-p=l-", we have

for all fixed x, as n 0

all tend to unity

and
 Diserete Probability Distributions | 73
Thercfore when n’0

when x = 0, 1, 2,.
Thus we observe that the Poisson distribution is a limiting case of binomial distribulion.

The Poisson distribution is a discrete distribution with a single paramcter m

It should be noted that

where m is called the parameter of the distribution and is the average number of
of random event x and e is the constant whose value is 2.7183. occurrences
The poisson distribution satisfies the two essential properties
i.e, P(X = x)>0 and P(X =x) =1.

§ 3.19. Form of the Poisson Distribution :

Like the binomial distribution, the variate of he Poisson
distribution is aalso a
discrete one ie., it makes only integral values. The probabilities of 0, 1, 2,.. successe
given by the successive terms of the expansion

+
2! 3!
where e = 2,7183 and m is constant called the parameter of the
distribution. We shall prove
that this m is the mcan of the distribution.
The expccted frequencies for diflerent succeses can be obtained
by multiplying
cach term of the above expansion by N ie. the total number of observations.
Following are some examples where Poisson distribution may
(1) Number of suicides reported in a particular city in 10 years. successfully be used.
(2) Number of deaths in a district in a unit of time viz, one month
by a rarc disease.
(3) The number of defective bolts per box of 50 bolts.
(4) Number of accidents in some unit of timc in some
(5) Number of wrong telephone calls received by a office country
in a day.
(6) The number of typing mistakes in a pagc.
(7) The number of radio active atoms decaying in time t.

S 3.20, Mean of the Poisson

distribution
The mean of the Poisson distribution is given by

0+
2!
74 | Satistics and Probability 1'heory

.me l+m+ " w .

Thus the mean of the Poisson distribution is m. and
uS
The variance of the Poison Distribtuion
on is given by
The variance of the Poisson
² =E(x') -fE(x)]² =E(r²)- m?

But x!
-*d,
dur
+ me .
(r-1)!
G
= m'e"e" + me ",e" = m²+ m
g = m + m- m = m
is also cqual to m.
Thus the variance of thc Poisson distributionof the Poisson distribution coinci
Important Note: The mean and variance
MGF about origin is given by
$3.21. MGF of Poisson Distribution :

Mo (1) = E()= P(X =x)

(x)!
mle-) ns
First four moments from MGF can be casily obtaincd. vith
Note. of a Poisson Distribution :
Probability Distributoieeon distribution is givcn oy
G
Probability distribution ..r....
X:
-m me "
P(X) Distribution :
Distribution Function of a Poisson
of a Poisson distribution is given by
tion
Distribution
m
F(r) = P(X'sr) = ) P(X =r) - ) r=0

Distribution :
Recurrence relation for Poisson

P(X = * +1) = P(X= x), where m is parameter. ve

x+1
Remark: The symbol has also bcen uscd for thc parametter.
Distribution :
§ 3.22. Fitting of the Poission
When in the practical situation m remains unknown tlhen it is estimated
the paramcter the Poisson
distribution
sample values. With that estimated valuc of
 Discrete Probability Distributions | 75
and expected or theorctical frequencies for thc differcnt variate valucs can be oblaincd by
using rccurrence formula as follows

f(r+1)
=) ...(1)
where f (*) is the expccted frcquency.
For x =0, S(0) = NP(S' = 0) = N = Nem (2)
From (1) and (2), all the frcqucncics can bc obtained by putting x = 0), 1, 2, 3,..
Mustrative Examples
Ex.l. Suppose on an average I house in 1000 in a certain district has a fire
during ayear. Ifthere are 2010 houses in that district, what is theprobability that exactly
5 houses have a fire during the year ?
Given e-0.1353 ). |Raj. BE (III Sem.) 03, G]

Sol. Here we have n= 2000, p = To00

oincide m= np = 2000 x
1000
Also, we have

32
P(S) = e22 - (0.1353) 2= 0.1353x
120
= 0,036 Ans.

Ex. 2. Suppose that a manufuctured producthas two defects per unit of product
inspected. Using Poison disribution, calculate the prububilities of finding a product
without any defect, 3 defects and 4 defects.
|Jodhpur BE III Sem, 04]
( Given e= 0,1353
Sol. Herc the averagc number of defccts i.e. mean =2

Now P(N = r) = e

P(0) =e2 = 0.1353 (Given)

Thus by using the recurrencc relation

P(Y =x+1)=P(X= *).

we havc P(1) =m P(0) = 2(0.1353) = 0.27
P(2) = P() = (0.27) = 0.27
ted by P(3) = P(2) =(0.27) =0.18
n is #
P(4) = P(3) = (0. 18) = 0.09, Ans.
 76| Statistics and Probability Theory
Ex. 3.A factory produces razor blades. The pprobubility of
1/ S00 . In 10,000 packets of 10 blades each, culculate its being det
the approximate
packet (a) having no defective blade (b) one defective blade (c) bro
(Given e-0.02 =0.9802) |Jodhpur def
BEectIl
ive b
Sol. Herc wc have N= I0, 000 n= 10 and p=

m=np=10x(s00)-0 02
Thus by using the recurrence relation
P(X = x + 1) = ). we have
I+
P(1) = n P(0) = 0.02 x 0.9802 = 0,019604

P(0)=0.02 =
P{2) = P() = x 0.19604 = 0.00019604
(a) Number of packets containing no defcctive = I0, 000 x 0,9802 = 9802
Numbcr of packets containing onc defcctive
= 10,000 x 0.019604 = 196 (Approximatc)
(c) Numbcr of packets containing two defcctive
10,000 x 0.00019604 I.96 = 2 (Approximate).

EX..4. A large numberuf observatiwns on u given solution whiclt contuined bas

were made by taking samples of ICC andl noting dlown the percetuge in euch sm
nu bucteri
Assuning the Poisson'ss distribution of 10% of the sample contained
the average number of bacterius present per C.C.
Sol. Let m be the number of bacteria pcr cc.
So F = m, Given 10% had zero bactcria

P(r= 0)= 0.1 -H=0.

So m= log, 10 = 2.3026

Ex. S. A manufucturer of cotter pins knows that 3"% of lhis product is defect
than 10 pins v
he sells cutter pins in boxes of 100 and guurantees that ot more
mwet the guara
defective, what is the appruximute probatility that a hox will fail to|Raj. BE III Sem
quality Sol. Given that n= l0)
Let p= probability of adofective pin = 5% = 0,05
Thercfore mcan number of defcctive pins in a box of 10)
= np = |00 x0,05 5
As the probability pis very small and the problcm is to find the approximate prob
we make usc of the Poisson distribution,

N=0.1,2,
 Discrete Probability Distributions | 77

Prob. that a box will fail to mect the guarantced quality is (.

e. x excecds 10)
P(X >10) = 1-P(X s10)

---) Ans.

Ex.6. Suppose that the nunber of telephone calls

exchange behveen 10 a.m and ll a.m say, X, is a random coming into a telephone
distribution with parameter 2 Similarly the number of callsvariable follows the Poisson
arriving between ll a.n.
and 12 noon say, X, has a Poisson
distribution with parameter 6 If X, and N, are
independent, what is the probability that more than S calls come beveen 10 a.m and 12
noon

Sol. We know lhat sum of two random

case of Poisson variables is also a random variablc in the
distribution So we consider
X= X, +X,
Therefore parameter m=m +m, m=2+6 =8
Hence thec probability of x calls come in between 10 am and 12 noon is
A given b
P(I =r)=
x
, x=0,1,2,.
Probability that more than 5 calls come in betwecn 10 a.m. and 12
noon is

P(X> $) =|- p(X s5) =l

5124096, 32768]
120
0.1912 0,8088.
Note. Use the table for- or calculator
78 ] Statistics and Probability Theory
Exercise IUL(6)
1. Fita Poission distribution to the following data which gives he number of ve
per square for 400 squarcs (given that e = 2.7183)
0 | 2 3 4 5 6 7 8 9 \0
No of cells (x):
(per squarec)
No of squares (/): 103 143 98 42 8 4 2 0 0 0

2. Fit a Poission distribution to the following data giving number of accident

during a period of 50 davs
No. of accidents (x) : 0 3 4
No. of days (): 21 18 7 3
3. f in a super markct the incoming ratc of customers is 10 customers per ho.
using Poisson distribution, find the probability of coming 4 customers in S mi

(e2.5=0,0821)
4. A fim produce artricles of which % are defective and packs them in lols of
articles cach, If a businessman purchases 100 lots, In how many of them be.
expect to have no defective, given that (e-0.5 0.6065)
5. The rcaction of a spccial type of injection to a person has a probability 0,001. Fi
probability wof
o reaction out of 2000 persons on:

(a) 3 persons
(6) more than 2 persons. ( Given e 0.1353) |Raj. BE n
6. Adriver has two taxis, which he hires out day by day. The number of demands
taxi on cach day is distributed as a Poisson variate with mean 1.5. Calculat
proportion of days on which
(a) ncither of the cars is used
2231)
(b) some demand is refused. (Given e-1.50
the prob
7. A telephone exchange receives on an average 4 calls per minute. Find
on the basis of Poisson distribution (m=4) of
(u) 2 or less calls pcr minute (6) upto 4 calls per minute
(c) more than 4 calls per minute. e 0.0183)
specimen of the
8. Red blood cells deficiency may be determined by examining a
contains on the aveg
under a nicroscope. Suppose a certain small fixed volume probability t
redcells for normal persons. Using Poisson distribution, obtain the
15 red cells.
spccimen from a normal person will contain less than
when therc is no miss
9. A manager accepts the work submitted by his typist only
per day of about 200 *
the work. A typist has to type on an average 20 letters
cach. Find the chance of hcr making a mistake
(a) If lcss than 1 % of the letters submittcd by her are rejccted
(6) 1C on 90% davs all the vork submittcd by her is accepted
 Discrete Probability Distributions | 79
10. Therc are on an average 4 accidents in a month in a factor Find the probab1lity of less
than 4 accidents in a particular month.
11. Assuming that the chance of atrallic accident in a day in a strcct of a city is 0.001.
On how many days out of a total of 000 davs can ve cxpcct :
(a) No accident
(6) more than 3 accidents, if there are 1000 such strects in the whole city
4 12. Afactory produces blades in packets of 10.The probability of a blade to be defcctive
is 0.2%. Find the number of packets baving two delective blades ina consignnment of
10,000 packcts.
13. What probability nodel is oppropriate to describe a situation where I00 misprints are
distributcd randomly throughout the 100 pages of abook ? For his modal, what is the
probability that a page observed at random will contain at lcast thrce mistakes ?
IHint : Use Poisson's
14
distribution| | Raj. B.E. II, 05]
In a lown l0 accidcnts took place in a span of 50 davs. Assuming that the number of
accidcnts per day follows the Poisson's distribution, find the probability that there will
be thrce or morc accidents in a day (Raj. B.E.II, 2000)
15. If thc chance that one of the ten telephoue lines are bust at any instant is 0.2. then
(a) What is the chance that 5 of the lnes arc busy ?
(b) Whal is the most probable nunber of busy lines and what is the probability of
this numbers?
(c) What is the probability that all the lines are busy ?
, 16. certain make of' automob1lc tyres get lat due to cxtemal causc, on the avceragc in
cvery 300O kn. Assuming that the occurence follow the Poisson's
law, deternine the
te
probability that:
(u) more than onc flat will occur in a S00 km run,
(b) no flat will occur in a trip of 6000 km
[Raj. B.E. II, 2001|
17 source of liquid is known to contain bacteria, with the nican nunber of bactcria per
cubic centimctre cqual to 3. Ten l c.c. test tubes are filled with liquid
Assuming that
Poisson distribution is applicable, find the probabilits that all tca test tubes
will I sshow
growth, that is contain at lcast onc bacteria cach
Jodhpur BE (EE), 2001
18 It is known that the probability of an itcm
produccd by a certain machinc will
defcctive is 0.05. If the produced itens arc sent to the market in packets of 20, Find
the number of packets containing atimost two defective itenns in a
1000 packets using (a) Binomial distribution (b) Poission distribution
consignment of
g
th
80 Statistics and Probubility Theory

ANSWERS
Exercise Ill (a)

1. c.(4). x0,1. 2.,...,3 2. 0344

3. 0.54 4, S 16
19/ 144
6. 1, 8, 28, 56, 70, 56, 28, 8, I:
Mean =4.06, S.D = |44
(a) 0.375
(b) 0.6875
8. 0.345, 0.246
9. (a) 90
(b) 9
10, 0.9186 11. 53/3125
12. (a) 0.2301 (b) 0341
13. (o) C,(0.2) (08 (b) 2,2"C, (0.2)' (0.8)
() Co (0.20
Exercise lli (5)
107, 141,93, 41, 13, 4, 1, 0, 0,0, 0
2. 20, I8, 8, 3, I
3. 0.1 (app)
4. 61
3. (a) 0,1804 (6) 0.3235
. (a) 0.2231
(6) 0.19126
. (a) 0.23
(6) 0.6283
(c) 0.3717

(a) 0,00005
10. 0.434
(b) 0.0000026
11. (a) 368 days
(b) 19 days
12, 2
17. 0.6 14. 03231
18. (a) 925 packets
(b) 920 packets