Analytics) KTU

160 (Unit-lV) (BtIsincss Statistics and

MBA Pirst Scmcstcr

particular X value (c.g., X = 56 scconds) becausc the BinomialDistrilbution

10.3.1. there arc only
valucs of X are not a sct of discrcte points. Rather, distribution take: pl.ace whCnoitcomcs. For
A binomial possible .
probabilities are defined as areas under a curve called Two mutually cxcusive may only reultin either
of a coin SCCess thcn
the Probability Density Function (PDF), Probabilities CXHmple, the lipping oneoutcornc I
we refer to aS fatlure
for a continuous random variable arc defincd on hcad or a tail. If othcr outcome
tcrm the
intervals such as P(55.5 < X < 56.5) or P(X < 56) or vC havc to 'B:TLGulli
P(X < 55). Figure 10.2 illustrates the differences rcferrcd io a,
distribution is also Mathcmatician Jacoh
between discrete and continuous random variables. This binomialafter the Swiss 10.3 show, the
Discrete Variable: Defincd at cach point Distribution
derived it. The figure
012345 Bernoulli who
cxample of coins.
Continuous Variable: Defined over an interval 0.5
Probability
54.5 55
53.5 54
53 0.25
Figure 10.2
(PDF) or Cumulative
Probability Density Function describes acontinuous 2
Distribution Function (CDF) PDF for a continuous 0
1
Number of heads
probability distribution. The
equation which shows
the 2 Heads
variable is an Probabilities of D, 1 and
random value of X. Figure 10.3:
F(x) at each possible
beight of the curve or 2 Heads
non-negative and the Probabilities of Getting 0, 1Probability
PDF must be Table:
Any continuous PDF must be 1. Number of Heads
entire 0.23
area under the
0.5
Definition continuous
cumulative distribution F(x) of a 0.25
The function f(x) is, 2
with density
random variable X
Probability
(x) =P(Xsx)=[ f()dt 10.3.1.1. Definition and

Function that of
for oo < X< oo.
consequence one can write
the two
probability of a successful rial is p and of
immediate If the the probability
As an
any trial is q (= l - p), dhen
in an experiment
results, failurein
F(b) F(a) and successful outcomes
Pla< X<b) = having X
is as follows:
dF(x) of n independent trials
f(x) = dx f(x)= "Cp'g"x
exists.
if the derivative the difference binomizl
examples that will clarify probability distribution is called the
Sorme
and continuous
variables are: This binonminal distribution is
betwecndiscrete air force sets
the probability distribution. The parameters n and p. If p and
Suppose that the Indian between a discrete distribution with
i)
qualification that all pilots
must weigh symmetrical, otherwise it is
The
be an
weight of a pilot wouldpilot's q are equal it is
55kg and 65kg. variable; since a nonsymmetrical.
cxample of a continuous value between 55kg and
any
weight could take on Notation for Binomial Probability
number successes that result from he
65kg. coin and counts the integer 1) r: The number of
one flips a
ii) Suppose any
number of tails could be could not,
binomial experiment.
of tails. The
value between 0 and
plus infinity. One
2) n: The number of trials in the binomial experiment.
tails. Therefore, the number of tails individual trial.
c.g.. get 2.5 variablc. 3) p: The probability of success of an
must be a discrete trial.
Probability Distribution 4) q: The probability of failure of an individual
Types of Continuous probability distributions are
continuous (This is equal to I -p.)
The following
illustratcd: Probability ii) Exponential 5) b(x; n, p): Binomial probability - the probability that
i)
Uniforn
Probability n-trial binomial experiment resulis in exactly ?
Distribution Distribution successcs, when the probability of success on a1
Probability iv) Student's individual trial is p.
iii) Normal Distribution
Distribution 6) "C,: The number of combinatiors of n things, taken I
vi) FDistribution at a time.
Chi-Square
Distribution
Pobability Distributions (Chapter 10) 161

Probability Table for a Binomial Distribution

Number of Probability Number of 2) Expand the binomial (q + pf. The number of terms in
Succcss Distribution Success Probabilty
Distribution
the cxpanded binomial is (n + 1).
"Cap'g= "C, p" a^ -n
3) Multiply cach term of the expanded binomial by N
(total frequcncy), to get the expected frequency in
"C,p'gl cach category.
"C,p'q n-2
"Ch-p4
3 "C,p' q
n-3
Example 1: Fit a binomial distribution to the following
"C,pq distribution of 156 samples:
10.3.1.2. Assumptions of Binomial
No. of Defective 7 6 |2
Distribution Items
) The number of observations is fixed.
No. of Samples 6 32 36 48 24 2

) Each trial has two Solution: To fit the binomial distribution to the data, we
mutually
outcomes, 1.e., success or failure. exclusive possible need the values of n, N, p and q.
3) Each trial is independent of other trials. X X Here,n =7, N= 156
2
Each trial has the constant 0 Mean np
p) or failure (= 1-p). probability of success (say
2 24 48
X=. X 546 =3.5
3 48 144 f 156
10.3.1.3. Characteristics of Binomial 4 36 144
Distribution 32 160 7xp=3.5
6 6 1
1) Adiscrete probability distribution which is based
36
Binomial Theoremn is called Binomial Distribution. on
7 1

2) Total number of trials 'n' and

probability of success 156
q=l
.

546
in each trial 'p' are the parameters of the
3) The mean and variance of a binomial
distribution.
distribution are Computations of Frequencies
'np' and 'npq' respectively.
4) Mean is always greater than its variance for No. of defective| f(x) = "Cp qr Expected
binomial frequency = Nx
distribution.
5) The binomial distribution is called f(x)
symmetric 1
0 P(0) =
1.22
distribution when p 128

6) The binomial distribution is called negatively skewed 1 P(l)=

7
128 8.53
1
distribution if p> 2 &if p< 2 then it is apositively 21
2 P(2)= 25.6
skewed distribution. 128
7) Binomial Distribution tends to Poisson Distribution if 35
'n' is very large and 'p' is very small. 3 P(3)=
128 42.65
8) Binomial Distribution tends to Normal Distribution if
'n' is very large; p and q are not small. 4 35
128 42.65

10.3.1.4. Applications of Binomial 5 P(5)=

21
25.6
Distribution 128

1) Itis used to find probability of getting 'x' success in 6 P(6)=

7

'n' independent Bernoulli trials. 128 8.53

2) It is also used in sampling, inspection plans, genetic
experiments, for finding control limits for p and np' 7
1,22
charts in statistical quality control.
Example 2: The bolts produced by a certain machine were
10.3.1.5!: Fitting a Binomial Distribution checked by examining samples of 12. The following table
When aibinomial distribution has to be fitted to an shows the distribution of 130 samples according to the
observed data, following is the procedure: number of defective items they contained:
) Detetmine mean of the given distribution and then
find the values of p and g. If one of these values is No. of defectives in sample of 12 01 234567Total
known, the other can be found out by the simple No. of samples 7619|35|3023]8|2| 130
relationship p + q= 1.
 162(Unit-IV) Statistics
MBA First Scmcster (Busincss
Analytics) AAKTU and
Fit a binomial
distribution and find the cxpecicd Example 4: The incidence of occupational disease in
Ircqucncics if the chance
Find thc mcan and of abolt bcing dcfcctive is l72. industry is such that the
workers have
20% anof
variancc of tlhe fitted distribution. suffering from it. What is the probability
disease?
that chance
out of six
Solution: Table: workers3 or more will contract the
No. of Computations of Frcquencies
defective Probabilities ("C,p" q) Expected Frequency = N Solution: The probability of a worker who is
P(0) =
x (x)

1.01
from the disease, i.e.,

p=100 5
20 1 sufeing
128
The probability of a worker who is not suffering
P(I)=
7.11 1 4 from the
2
P(2)= 21
disease, i.e., q=1-p=1- 5

128 21.33
The probability of 4 or more, i.e., 3, 4, 5 or 6 will
3 P(3)= 35
128 35.55
disease is given by, contract
P[X>3] =P[X= 3] +P[X= 4] +P[X= S] +PX 6
4
P(4)= 128 35.55
21
128 21.33
6
P(6)=
128 7.11
6! 64 6! 16
X
P(7)= 3!x3! 15625 4!x 2! 15625
128 1.01
6! 4 6! 1
Mean = np=7x=3.5 and Variance = 5!x 1! 15625
+
2 npq =7x 1/2 x 6!x 0! 15625
1/2 = 1.75
3x2
Example 3: The mean and variance of a binomial
X are 4 and 2 variable
respectively.
takes values greater than 3.
Find the probability that X 6x

=20x/_64 ) 16 4
Solution: Given mean np=4 +15x 1
and
Variance npq =2
Dividing (2) by (1), then g =
.....(1)
, ..(2)
15625)
+
240 T5623J+6x15625)15625)
+
24 1

1545
15625 15625 Js625
i: p=l-q=1-=.
2 2 15625)
=0.098

Putting the value of p in equation (1), nx- =4’n =8

10.3.2. Poisson Distribution
Now P(X > 3) = P(X = 4) + P(X = 5) + P(X = 6)+ P(X= Poisson distribution
was developed by the French
7) + P(X= 8) mathematician Simeon Denis Poisson in 1837 (1781-1840).
Poisson distribution arises when you count a numbe
events across time or over an area. Poisson
discrete distribution is a
random variable distribution that
expresses probability of the given number of events in a fixed
interval of time, if these events occur with a
rate and independent of the time since the lastknown average
event. Poisson
distribution can also be used for the number of events in ouio
specified intervals such as distance, area or
8 8 8! 8! 8! 1 volume.
4x 4! 5% 3! 6x 2! 7%1! 8x 0! 256 Poisson distribution can be used as an
Binomial distribution in the case of alternat1ve
8x7x6x5,8x7x6,8X7,8 1 binomial distribution the value of n very largë sámples.
is finite. Byt there ard
4x3x2 3x2 256 few problems when it is not possible to
this case binomial distribution is estimate 'n.
{(2x7x5)+ (8x7) +(4x) +8+1} 163 = 0.636 inapplicable
cannot expand (p + ), if n is not known. Inbecause Oe
Such cases
256 256
Poisson distribution is applicable.
robability Distributions (Chapter 10) 163

Examplesof Poisson
Thenumber of Distribution successcs that occur in a
) tclephone calls reccived at a 3) x: The actual number of
Switch board per minute
particular
during certain hour
a
of the
specificd region.
m): The Poisson probability that exactlythex
day. 4) P(x;
)
The number of
accidents pcr ycar in a district of SICCCSscs occur in a Poisson cxperiment, when
Maharashtra. mcan number of succcssCs 0S J.
3) The number of
trucks passing a certain
minutc. point pcr 10.3.2.2. Assumptions of Poisson
The number of persons bom
4)
in a city. deaf and dumb per Distribution
The number of
ycar ) The number of events is discrete in a given interval,
5) typing
Thenumber of printing
errors per page. say [0, t].
6) errors per page in abook. 2) Thc probability is approximately proportional to the
7) The number of defective blades in a length of the interval for an event which may occur in
pack of 1000.
pifference between Binomial a short interval [t, t + AtJ.
and Poisson Distribution 3) The occurrences of events are independent in non
Basisfor Binomial
Poisson Distribution overlapping intervals.
Comparison Distribution 4) The probability of twoevents is negligible in a short
Meaning Binomial distribution
|is one in which the Poisson distribution
interval [t, t + At).
gives the count of
probability of
repeated number of independent eventswith a
Occur randomly
10.3.2.3. Characteristics of Poisson
trials are studied.
given period of time. Distribution
Nature Biparametric. 1) The outcomes that occur in the result of the
Number of Fixed Uniparametric.
Infinite experiment can be classified as successes or failures.
trials 2) The average number of successes () which occur in a
Success Constant probability. Infinitesimal chance of specified region is known.
Outcomes |Only two possible
|succesS. The probability that a success willoccur is directly
Unlimited number of related to the size of the region.
outcomes, i.e. success possible outcomes. 4) The probability that a success will occur in an
or failure.
Mean and Mean > Variance extremely small region is almost zero.
Mean = Variance 5) It is a discrete probability distribution where the
Variance
random variable x assumes the infinite set of values 0,
Example Coin tossing Printing mistakes/page 1,2....
experiment. of alarge book. 6) i) Mean (m)
ii) Standard
10.3.2.1. Definition and
parameter ofi) Variance (")Deviation (G) =
Function
Probability the
Vm
distribution,
Itis defined by the probability function: iv) Skewness =v) Kurtosis
emX
f(x) = (Peakedness) = 1
x! m
7) The mode of Poisson distribution is that value of x
for x (no. of successes) = 0, 1, 2, 3, .......where m is a
fixed positive number. which occurs with largest probability. It may have
either one or two modes. If 'm' is not an integer, the
e= 2.71828 (app.) mode in the integral value between (m- 1)
If, however, 'm' is an integer, and (m).
Ihe successive terms of the distribution for different then there are two
modes which are (m-1) and (m).
values of x are as folows:
8) If x and y are two
|X:0 1 3 Total independent Poisson variates with
1 parameters mand m, respectively, then their sum (x
+ y) is also a Poisson
(x-m
e xm'ex m' (as the total variate with parameter m, + m,.
Xm
34
probability must be 9) The first, second, and
2! third row moments are
unity)
respectively (m), (m´+ m), (m' +3m' +m).
Notation for Poisson Probability 10.3.2.4. Applications
The talk about the of
following notation is helpful, when we Distribution Poisson
Poisson
1) e:
distribution. approximately 2.71828. 1) To count the number of
defects of an item used in
Constant equal to quality control' statistics.
(Actually, e is the base of the natural logarithm 2) Tocount the number of
System.) 3) To count the number ofgerms in a substance,
radioactive substance in a particles discharged from a
2 successes that occur in a
u: The mean number of
nuclear reactor.
specified region.
 Semester(B
MBAFirst
164 (Unit-IV) defective
Noneare ex40
4) To count the number of casualtics with the customer ) ee=0.018
of an P(X =0)= 0!
insurance
S) In qucuing company.
problems to count thc number of arrivals
P(O)= 0.018
of custonmers at a billing
counter. defective
6) Number of traffic arivals such as acroplancs at 5 geysersare
airports, ships at harbours, tnucks at the gates, ctc. 2) ex45 0.018x1024 18.432
P(X =5) =: 5!
n In dctermining the total number of deaths carricd by a
specific discasc (and not cpidcmic), in a particular
locality or a particular (spccificd) period of time. P(5)= 0.154
5x4x3x2 120 =0.154
8) The number of typing crrors Poisson frequency
material.
9) In problems dealing with the
pcr page in

manufactured products with the probability

yP

inspection O
Example 6: In a
frequency
corresponding to 4 successes is 5/3
to 5 successes. Find
frequency corresponding distribution
distribuiithmoen,es thehe
that any
One piece is defective is verv small and the lots are and standard deviation of
the meanN
very large. = e m
10) To model the distribution of the Solution: In Poisson Distribution f(x)
number of persons
Joning a queue to receive a service or purchase
X!
product. or a Where., x= No. of successes, m = Mean
e"m(x+1)
f(x + 1) =
10.3.2.5. Fitting a Poisson Distribution (x+1)!
The process of fitting a Poisson
First one has to find the value. distribution
of m.
is very simple. f(x+1) m
ie., the average f(x) (x+1)
oCcurrence, and calculate the frequency of 0
The other frequencies can be successes.
f(x)e "m
calculated as follows: f(x + 1) =
m

e m(x+) (x+1)
x! and f(x +) =: When x=4,
(x+1)!
f(x +1) m+l) x! m (S)=(4) "(1)
f(x) (x+1)! m' (x+1)
Since f(4) =
f(x +1) = m . . . (2)
(x+1)f(*)
3
When, x=0’f(1)= mx f(0); From equation (1) and (2),
When, x= 5
1’f(2) =.m?
and so On.
f): When, x=2’f(3) =f(2):
3
(5)=(5) 5
m=3
3

This recurring Standard Deviation = m=3 =1.732

for fitting the relation provides a very convenient method
Poisson distribution. The only Example 7: Suppose that a
isf(0) =e, where m isprobability
we need to know
the manufacturer
per unit of product in find 3 defects
parameter of the Poisson the only inspection. Using Poisson
(the total eachdistribution.
By multiplying distribution,
without any calculate the probabilities
distribution, frequency)
we get the expectedprobability of the
N
Poisson defect, 4 defects and 5 of finding aproducl defects. (Given e=
probabilities. frequencies for respective 0.0498)
Example 5: If 4% of the electric geyserS Solution: Average number of defects,
a company are f(x) = emx mn=3
the probability indefective, use poisson manufactured by X! -,X=0,1,2,...
i) None is
a sample of 100 distribution
geysers when:
to find
defective ii) 5geysers are f (0) =
(Given: e=0.018) defective e=0.0498; m
f(1) = f(O) x m= 0.0498 X
3=0.15;
Solution: Let p=
probability of a defective geyser f(2) =f(1) × 2 m

= 4% =0.04 0.225; =0.15 x1.5 f(3)) =f(2) x 3 = 0.225x

We are given n = 100, p =
100
3
l=0.225;
0.04; .. f(4) =f(3)
0.04 = 4 Mean (m) = × 4 =
0.225 x 4 f(5) =f(4) x 5 : = 0.17x
The Poisson
np:100 x =0.17;
distribution
e m'
is given as: 3
P(X = x)= 5 =0.102
x! Hence, the i
that a no defeet