PROBABILITY DISTRIBUTIONS 483

ProbabWty Ma•• Fanction,

12 A probarulity ruaction that specifies the probability that any single
'Value of discrete random variable will occur is 'called a probability mass
function (abbreviated as p.m.f.)._ If /(x) is the probability mass function of
the random variable X, thenf(x)=P(X=x) has the following properties:
ProbabilitY Distributions (I) f(x);;>O for all values of X; and
(Ii) l:/(x)= L
In this chapter the basic concepts of proba~[Link] d'istri>u~~ ant Cumulative Mass Fan~o'n
discllsSOd> JD a probability dis~bution, _the vanables ar_,, dietribolcd If X is •a discrete random variable with p.m.f. /(.x), its cumulative
accordinl to some definite probability function. We ~all disCUSS some or
the most important probability distributions in t~ ~~P~· These
mass function (abbreviated as e.m.f.) specifies the probability that an
observed value of X will be 110 greater than x. That is, if F(x) is a e.m.f.
distn1>utions from their histoti~ in~t as. well as ~tnns•c ~rtance and/(x) is a p.m.f., then F(x)"7 P(X<x) f(X<x).
occupy a place of great prominence 1n bu$mess dccJSIOO.-JDaJciq.·. Some
of the .concepts that will facilitate understanding of the topic are given ,Continuous Probability F1111ction
below: • A probabiJity function for a continuous random variable is called
Bendom Variable. A random variable is a variable wliicb. takes. :a continuous probability function sinde the domain of the function is
tpee?ed value.1 with specified probabilities. The probabilities are speclned. -continuous. \
by the way in which the random experiment is conducted and the way in Probability Density Fanction
which the random variable is defined and observed on the random
apcriment. 'W_e [Link] capital letters to denote a random variable and For a continuous random variable, the corresponding function f(x) is
the CQ..~nding small letters to represent any specific value of the random -called a pr<;>bability density function (abbreviated as p.d.f.). Unlike a p.m.f.
variable. • a p.d.f. does not specify probabilities for specific individual values of the
random variable. '
Probability Function. If the· function permits m to compute· the
~ for 8!1Y even_t t~t is defined in !~ of value of the random Cumulative Density Function
variab_le, than this func!,ion IS call~ a probab1bty function. J~t as there Corresponding to the cumulative mass function of• discrete random
are ~ .and ~ntinuo~ random variables, so there are discrete and variable the cumulative density function (a'bbreviated -as c.d.f.) of a conti-
,ara1hnP0111 ,robability functions. To emphasu.e tJm. distinction, we shaO nuous random variable specifies the probability that an observed value of
draw tbe rffllll'lm as follows : • X will be no greater than X. •
·Espected Value and Variance
PROBABILITY FUNCTION
The ~ro~abi!ity distributi~n provides a model for the theoretical
fre9uency distribution of a random variable and hence must possess a mean,
variance and other descriptive measures associated with the theoretical
_population .which it represents. The average value of a random variable
as· called the ex~ied value of th«? random variable.
DISCRETE PROBABILITY CONTINUOUS. PROBABILTY Let X b~ a discrete random variable with probability distribution,
[Link] FUNCTION •
P(X), then the expected value E(X) is give~~y
, E(X)=EX.P(X) •
where the element_s are summed over all values of the. random variable X .
1
In other '!ords, if a discrete random variabl~ X has possible values
~[Link],,,
r,ASS t:UNTION
II CUMULATIVE
MASS FUNCTION
• PR08ABIUTY ' f CUMULATIVE
OENSfTY F\MCTION • (DENSITY FUNC~
.Xi, Xs,, .. , x., with corresponding probabilities J> (xJ, P(xJ, ... , P(x.) then
the expected value E(X) is defined as
E(X)=x1 P(xJ+xJ>(xJ4-.~-+xJ>(X.)=,a
' ' Thus th~ expected value of random variable Xis merely the arithmetic
Discrete Probability Function mean which may .be denoted .by f'..
4
A probability function for a discrete random variable is called a
discrete probability function since the domain of the functioa is dilcrett.
 - ~ - :i$iL": ~- -._•-.
BUSINESS STATISTICS
PROBABILITY DISTRIBUTIONS 485
484
SimiJarlY, the variance of tbt probability distribution• of thq rando1111 Solution. The probability distribution of net profit (X) of the new product in the fint year, is given
variable x is defined as the expeci':'1 value of tho sum• of the squared to be
deviations of the values of X from. thell' mean. -I
Thus the variance of the discrete· random variable Xis given by ,· ,,,t Profit (in million Rs.) X 3
Var. (X)=a'=E[X-E(X )Jl=l:[X-E(X)]"P( X) Probability P(X) [Link] 0.2S 0.60
=E(XS>-[E(X)]1 • Therefore, expected value of profit is given by
The standard deviation, a, is the square root of the variance-. E(X) = :EX P(X)
Prope1 ties 0£ Espectecl Value and Variance = 3 x 0.15 + I x 0.25 - I x 0.60
[Link] are several important properties of expected value and variance which = 0.10 million Rs. - Rs. 1,00,000.
E(X-) = Lr- P(X)
allow computational shortCUts : •
1. The expected value of a constant c is equal to the constant. = 9 x 0.15 + I x 0'25 + I x 0.6 =- 2.20 million Rs.
Var (X) = E(X-) - [E(X)P = 2.20 - (0.10:f = 2.19
E(c)=c s.d. =a= ✓2.19 • 1.48 (Million Rs.)
. 2. The expected value of the pcoduct of a constant c and. a r~dom
vanable Xis equal to the constant times the expected value of the random Binomial Distribution
variable. The Binomial distribution is also known as the outcome of Bernoulli proc-
E(cX)='cE(X) ess and is associated with the name of Jacob Bernoulli. A Bernoulli process is
3. The expected value of the sum of a random variable X and a a random process in which
constant c is the sum of the expected value of the random. variable and the (a) The process is performed under the same conditions for a fixed and
constant. 11 finite number of trails, say, n.
E(X+c)=E(X)+c (b) Each trial is independent of other trials, i.e., the probability of an out-
· . 4.1:he expected value of the product of two independent random come for any particular trial is not influenced by the outcomes of the other
vanables 1s equal to the product of their individual expected values. trials.
E(XY)=E(X)E(Y) . ; ,:ri (c) Each trial has two muturally exclusive possible outcomes, such as
"success" or "failure", "good" or "defective", '6yes" or ..no", "hit" or "miss".
, . b5- TJie expected value of the sum of the two independent random and so on. The outcomes are usually called success and failure for convenience.
vana 1es 1s equal to the sum of their individual expected values.
(d) The probability of success,p, remains constant from trial to trial (so is
E(X+Y)=E(X)+E (Y) • the probability of failure q, where q = 1 - p).
. ~ The variance of the product of a constant and. a random variable X These conditions are statisfied if we toss a coin, say, five times. Suppose
18 equ to the constant squared times th~ variance of the random variable X. we are interested in finding the probability of obtaining exactly two heads.
Var (cX)=ez Var (X) • Let us designate head as a success and tail as a failure with corresponding
"d~m ~f two 1D
probabilities p and q respectively.
[Link]
The·
sumvariance of the
of their indi • dependent random· variable&
rence of tw · d vi vanancc;s. Also the varianee of the dilfo- Suppose that one of the sequence of outcomes of five tosses ofa fair coin
varianccs. 0 in ependent ran<t;om variables equal the sum of their individual showing two heads is :
HTHTI'
. Var (X+ Y)~ Var(X)+ Var(Y). Var(X-Y). The probability of this specific sequence of outcome is found by means of
~ t. Aoil compan timatea the net profit on a new product it ii a multiplication rule of probability and is given by
J.a1;1o~h1n1 to l,t lb. 3,000 000 d · .Y Cl
if 1t •• rJoderatcJy auccesaful a To, the fint year if it i1 •111ccct1ful~ _. Rs 1 000 OOC, = p2ql
t~:ri~•°.
e~t!l! • 1',
the folloVviog •:robabi~ltr:.~- 1.:_0,000· if it i• •Ulllu~eairu'I'. Tb•
mod dcrately •~•lfut,: . °and·~ year
pqpqq
Althoj_gh the resulting probability of obtaining Jhe specific sequence
protpect1 for the product.,.
8 1oe an standard devia1ion ef 0fkat
25• naucceaaful : 0·60. What ar• tla• of outcoFs. in the order shown, we are not interested in the order of
· year net profit for the prochaei t. ._
occurrence"'of the successes and failures. Rather, we are interested in the
probability of the occurrence of exactly two successes out of five tosses
of a coin. In addition to the sequence shown above ( call it sequence)
486 BUSINESS STATISTICS PROBABIL ITY DISTRIBUTIONS 487
number I) twO auccesses ancl'.tbrCO fail- could also occur in Uly on~ of (n - x) failures inn Bernoulli trials. The binomial distnbutio n has been extensive ly
the additional IICClueoces shown as follows. Each-of the sequences has tho tabulated for different values of x and n (See Appendix!.
samo probability of ()CCUl'rinS, p'r/. Number ofsuccesses x I Probabili ty /(x)
s,q/,el!" ]iumb,r I s,quence I p,oba/,Utty 0
1
nCoPoq" - o = q"
nCJJ'tt' - 1
= nq11-•p
HH1Tl' ~~ n(n-1)
2
3
4
s I
HTITH
'JTI'HH
TI'HHT
..
~::.
...
2 If C1plq11-l =---"----q
2
•-l p l
.. rtt
. X n Cx[T'q" - ·'
10 I : \
A single sample of five tosses will yield only ono sequence of
nccesses and failures. The qu~tion then to be answered is : What· is· n I nC,,P"q" - •= p"
tho probability o f ~ • scquenco nwnber 1 or seq~encc numbor.2 ••• ,..
or aeq11CGce· number 10?. For pnding the answer, lhe addition rule t,f The binomial distri~ution satisfies the two essential properties of probabilit y
probability is used to calculato the sum of tho individual probabili ti~. distribution, viz.,
To get this we multiply p'q' by__lO~_I.~., 10 yq'. - • (i)/ (x) ~ 0 ; and (ii)"£/ (x) = 1.
Herc p=0·S an~ q-:-0·S
For (i), this follows from the fact that both n andp are positive and hence
Therefore, the answer is
lO(O·~(O ·S)'=JOX0 °2SX0°12 S=0·0312 5 nC,., y, q,,_x all are positive. Conseque ntly /(x) ~ 0.
. As tho size of the tosses increases, it· becomes more and more difficult For (ii), we know that binomial expansion of
to ~--athe number o~ sequences. Ail easy method of counting them is\
rcq~ (q + p)" = q" + "C,q"-•p + "C2q"-2P1 + ·······+ ·c_,q"-..P, + ...... + P"
,
. .w~ know that the .number of combinations of n things taken x at a
1 Therefore "£/{x) = l:"C., p'q" -., = (q + p)" = (1 - p + p)" = 1.
tiJJ}ellglV ~by •
The binomial distirbution is a family of distributio ns since each different
nl"
•C•= x l (n-x)l value of n or p specifies a different distribution. In this distributi on n and p are
In ou, example, n=S. x=2. called parameter s. Regardles s of the value of n, the distributio n is symmetri cal
when p = 0.5. For small values of n, when p is greater than 0.5, the distributi on
Then C1 2 ~ ; l -10.
1
is asymmetr ical, with the peak occurring to the right of centre, i.e., it is a
Therefore, the general model for specifying the prob~bili9', o_f negatively skewed distribution and when p is less than 0.5, the distributi on is
obtaining exactly x successes in a given number 01' n Bernoulli -trials. 11 asymmetr ical with the peak occurring to the left of the centre, i.e.. it is a
given by positively skewed distribution.
j{x)=P[X- x]=•C..ps q"-:-• for x=0.-1, 2, ....·.:,-n• na10 •40 .40
.35 p.0.05 .35
where n=10 .35
p=tho probability of a:tuce;es s on i'.30 ~-30 p:0.1 .30
n=10
Bernoulli trial i
1
..c .25 .25 p=0.09
.25
n=the number of Bemo trials .§ .20 1 .20 .20
x=thc number of successes in n trials. ct .15 a. .15
.10 .10
a: .15
This formula. for the probability distribution of.. the number ·of .05 .05 .10
succcsscs in series of Bernoulli trials is called the Binomial probability 0~~~~~~~ I I I I I!••••• .05
distribuf;ion. -It gives the probability of pbtain~g ~ l y x successes ana 0123456 78910 O 0123456 78910 0 1234567 8910
X X X
The diagram given above for n = 10 andp == 0.5, 0.1 and 0.9 makes this
distinction clear.
 \
I
BUSINESS STATISTICS PROBABILITY DISTRIBUTIONS I
489
488
Mean and Variance of Binomial Dlstlrbu tlon Therefore 40
np=40o rn-=-=4 00
The Mean. The mean of binomial random variable X, denoted by µ or p
E(X), is the theoretical expected number of successes in n trials. Hence for the given question, n • 400, p • 0.1 and q = 0.9.
lllustratlo n 3. Assuming that half the population is vegetarian so that choice of an
" individual being
µ = E(X) =: 1:x f(x) a vegetarian is 1/2. Assuming that 100 investigato rs can take a sample of 10
xsO individuals to see whether they are vegetarians, how many investigatores would you expect
i.e.. the mean of Xis the sum of the products of the values that X can assume to
report that three people or less were vegetarians 7
multiplied by their respective probabilities. Solution. Probability of a person being vegetarian, i.e.. p = I
µ = E(X) = [Link]{x) = I.x"Cx p'q"-x I 1 2
q=l-p =I--= -
2 2
"
~
n'
• .I ,r-.r The probablity that three people or less are vegetarian is given by
-
-
/.Jx ( ) p q
x=o x! n-x ! P[X S 3] = P(0) + P(l) + P(2) + P(3)
~ n(n-1)! x .. -x
= q'o+ ('oC,)q'p + c•oc~)q',r+ ('OCJtl'P"
= ~x p q 0
x(x- l)!(n -x)! = (.!..\'
i)
+to(! \'(!\+ ~ (!\' (!Y+ l0x9x8 (!\' (!\'
i) i) 2 x I i) 2) 3 x 2 x I 2) 2)
= '[Link] (n-1)! Px-•q"-x
l\'o (1\'o
xsl {x-l)!(n -x)! = ( 2) +10 2) +45 (I\'o (l\'o
2) +120 2)
= npr.-•c_,_,Jr'q-' = np(q+p),t-1 = np (-: (q + p)"-' = 11 0
= (-
1)' [ I + IO + 45 + 120) 176
= --
Thus the mean of the binomial distribution is np. 2 1024
The Variance. The variance of the binomial random variable :. The number of investitato rs to report that three or less people are vegetarian s
X measure given by
is
the variation of the binomial distn"bution and is given by ' 176
o2 = E(X2) -µ2 = LX2 " f(x)- µ 2 1024
X 100 .. 17.2
)
x=O Hence 17 investigators would report that 3 or less people arc vegetarians.
Here µ= np Illustration 4. The incidence of occupation al disease in an industry is such that the
workers have a 20% chance of suffering from it What
and "£.x:/(x) = "£.[x (x - 1) + x] "C.r p'qtt-·• 4 or more will contract disease ?
is the probability that out of six workers
(MBA. Lucknow. /990; MBA. DU, 2000)
= "i.x (x - 1Y'Cxp•q-r+"i.x'Cxp•qtt-·r
Solution. The probability of a worker of suffering from disc:asc:, i.e., p = 20 =
I
= n (n - l)p2 (q + py,- 2
+ np The probability of a worker not suffering from disease 100 5
= n (n - l)p2 + np [-: (q + p)"-2 = 11 I 4
i.e. q = 1 - p = I - S = S
0 2 = n (n - l)p2 + np - (np)2 The probability of 4 or more. i.e.. 4, 5 or 6 will contract disease is given by
= np [(n - 1)p + 1 - np] P[n4] = P[4) + P[5) + P[6)
= np [1 - p] = npq [Since p + q = l] = (6C.) (~J (:J+ ('Cs) (~JG)+ (½J
Thus the standard deviation of the binomial distribution is .Jnpq and
variance = npq. =IS (~5 (:} (~J (~)+ (½j
+ 6
IS x 16 6·x4 I
lllustntlon 2. The mean of a binomial distribution is 40 and standard deviation 6. Calculate = 1S625 + 1562S + 15625
n, p and q. [MBA, DU. /998] 1 265 53
Solution. The mean of binomial distribution is given by np and standard deviation = 15625 [i40 + 24 +l] = 15625 = 312S = [Link] 696 •
by
Illustration S. Assume that on an average one telephone number out of fifteen is busy.
Jnpq.
What is the probability that if six randomly selected telephone numben arc called
Since Jnpq = 6; npq = 36 and np =- 40 (a) not more than three will be busy '!
• 36 (b) at least three of them will be busy '!
Therefore 40q = 36 or q =
40 = 0.9 Solution. p = probability that a telephone number is busy =
I
p = I- q = I- 0.9 = 0.1
15
Since p = 0.1 and q = 0.9 and therefore, npq = 36
q = \- p = 1 - - 1 = -14· and n = 6.
15 15'
 BUSINESS STATl~TICS c ' ~
491. ·····\.
490 PROBAB ILITY DISTRIB UTIONS
(a) The probability that out of six randomly sel~ted telephone numbers
nol mo"! than
When x = 1,
thtec numbers arc busy is given by 2
P[X s 3) • P(0) + P(I) + P(2) + P(3)
1
( ~ )J(_!_1s)3 /( 2)=P .!:_!/( l)=(p) n(n-1) f(O)
14 6
=(15) +(
6
14 s l
c;(15) (IS)+(
6
ez(JS) (15I
14 4 )2
+(
6
c, 1s Whenx = 2,
q 2 q 2!
3
=( .'sr [(14)6 +6(14)5 + 15(14)" +20(14) ] pn-1
/(3)= --/(2 )= ( P )
(
3nn-1}(n-2) /(0)
q 3 q 3!
4
3
3 2 and so on.
= (t ) [(14) +6(14) + 15(14)+2 0]
(1s)6 This formula provides us a very convenient method for fitting the binomial
=
2744 [2744+11 76+210+ 20] distribution. The only probability we need to calculate is/(0) which is equal to
(15)6 I
qn, where q can be estimated from the given data.
_ 2744x415 0 = 11387600 = 0 _9997 Illustrati on 6. The screws produced by a cenain machine were checked
by examinin g
n of 128 samples
- 11390625 11390625 number of defective s in a sample of 12. The following table shows the distributio
(b) Probability that at least three telephone numbers arc busy is givcny by according to the number of defective items they contained :
No. of defective s
P[X ~ 3) = P(3) + P(4) + P(5) + P(6) 7 Total
2 3 4 5 6
(::rc~r (::rc~r
in a sample of 12 : 0 1
6
=( c3
6
+( c.. +(
6
~(::)L'sr + (
6
~)( .'sr [Link] ples
being (c1)
Fit a binomial
defective is ½.
7 6 19 35 30 23 7 I 128
distribution and find the expected frequencies if the chance of machine
= .0051 (b) Find the mean and standard deviation of lhe fitted distribution.
(MBA, DU, 1993)
Fitting a Binomial Distribution. When a binomial distribution is to be Solution. (a) The probability of a defective screw = ½.
fitted to observe data the following procedure is adopted : , p = Yi. q .. ½. N • 128.
1. Determine the values ofp and q. If one of these values is known
the Since there are 8 terms, therefore, ,i = 7.
other can be found out by the simple relationship p = (1 - q) and q = (1 - p).
of:
The probabili ty that7 0, I, 2, ............. , 7 will be deffective is given by expansion
When p and q are equal, the distribution is symmetrical. Then p and q may be
1 ½)4
(½ + ½)7 = (½) + ('C,XY1)6(½) + (7C2)(½>5(½) + ('Cl)(½) (½)3 + ('C4X½>3(
2 4
interchanged without altering the value of any term, conseq uently terms + <'cs)(½)2(½)s+ <7c6)(½)6(½)+ c7c1)(½)7
equidistant from the two ends of the series are equal. Ifp and q are unequal the
7
= (½) [1 + 7 + 21 + 35 + 35 + 21 + 7 + I}
distribution is skewed. Ifpis less than 0.5, the distribution is positively skewed In order to obtain the expected frequencies we shall multiply each term by
N. Le., 128.
and when p is more than 0.5, the distribution is negatively skewed.
1
1
-(1 + 1 +21 +35+35+ 21 + 1 + 1)
2. Expand the bin~mial (q +[Link] e power n is equal to one less than the 128(!+ !) = 128 x -128
2 2
number of terms in the expanded binomial. Thus when n = 2 there will be three Thus the expected frequencies are :
tem1S in the binomial. Similarly, when n = 4 there will be five
terms. X O 1 2 3 4 5 6 7
3. Multiply each term of the expanded binomial by N (the total frequency), f : 1 7 21 35 3S 21 7 1
in order to obtain the expected frequency in each category. (b) Mean and standard deviation of the fitted distribution
It is convenient to use the following recurrence relation for fitting of Mean of the binomial distribution is np and standard d~iation is Jnpq ·
n = 1, P = ½. q =½
binomial distnbution. Here
Mean = np 1 x ½ = 3.5
f(x)= P[X= x] = "Cxp·fq" - x
f (x + 1) = P[X= X + 1] = "C.r+Ipf+ 'qn-:c-1 Standard deviation = Jnpq = J1 x ½x ½= J1.1s = 1.32.
f(x+l) p n-1
.. . . - is the Poisson
f(x) x+ 1 =-;; Poisson Distribution
• A second important discrete probabthty _dtSlnbubon ho ublished
P
p n-x ~•stribution, named after the French mathematictan S. [Link] w
f ( x+l ) =-- f(x) •ts derivation in 183 7
q x+l
Th •
• of the Poisson the :
di stnibution arc as .follows ce of
When x= 0, e characteristics
p n-0
/(1) = - -
p
/(0) = -n f (0) I. The occurrence of the events is independent. That~ °:.i,"; of a
q O+l q an event in an interval of space or time has no efft:et 00 • ~ .
• the same, or any other tntir;olll,
second 0 ccun-ence of the event m
492
BUSINESS STATISTICS
2. Theoretically, an infinite number of occurrences of the event must be
. .

PROBABILITY DISTRIBUTIONS
The Variance. The variance of the Poisson distribution is given by
2

' ,,,
493

'
•'
possbile in the interval. . . cr = E(X 2 ) - [E(X)]2 = E(X 2) - m 1
3. The probability of single occurrence of the event in a given interval 1s
proportional to the length of the interval. . . .. e_,,,mx
4. In any infinitesimal (extremely small) portion of interval, the probab1hty -ml
= Ul
x!
of two or more occurrences of the event is negligible.
Poisson distribution differs from the binomial distribution in two important e_,,,mx [x(x- 1) +x] x
But :Exl =e-r. m
respects. . . . . . x! X !
(a) Rather than consisting of discrete tnals, the d1stnbut1on operates
continuously over some given amount of time, distance, area, etc. - 2 m mx-1
x-2
(b) Rather than producing a sequence of successes and failures, the - m e-r. (x-2)! +me-r. (x-1) ! - m l
distribution produces successes, which occur at random points in the specified
time, distance, area. These successes are commonly referred to as 2
= m e- • e"' + me- • e"' - m2
'occurrences'. = m2+m-m2 =m
The Poisson distribution may be used to approximate binomial distribution
when n is large and p is small and, therefore, is regarded as the limit of the Thus the variance of the Poisson distribution is also equal tom.
binomial distribution : The Poisson distribution is completely defined by the parameter m and is
The Poisson distribution is given by
e-mx positively skewed. This positive skewness is typical of the Poisson distribution,
f(x)=P(X=x) = - - ; x=0,1,2 ... indicating that, with extremely small probability, there is the possibility that
x!
where m is called the parameter of the distribution and is the average number distribution will produce an indefinitely large number of occurrences in a
1of occurrences of random event, x is the number of occurrences of the random segment of time or spa~e, even though the mean rate of occurrences may be
event and e is the constant whose value is 2.7183. quite small. As m increases the distribution shift to the right. This is illustrated
The Poisson distribution satisfies the two essential properties, i.e.,/ (x) ~ below in the diagram for 4 values of m from m = 0.3 to m = 4.0.
0 and 'f./(x) = 1.
THE POISSON DISTRIBUTION
The Poisson distribution has been extensively tabulated (see the Appendix). 1.00 I
It has many applications in business and has been widely used in management
science and operations research. The following are some of the examples which
may be analysed with the use of this distribution :
(a) the demand for a product, 0.75
(b) typographical errors occurring on the pages of a book, 1
1\ym = 0 -3
(c) the occurrence of accident in a facotory, I \
(d) the arrival pattern in a departmental store,
(e) the oc~urrence of flaws in a bolt in a factory, and
I
I
' '\I
0.50
(f) the arnval of calls at a switch board.
Mean and Variance of the Poisson Distribution
' '
I~\ I
The Mean. The mean of the Poisson distribution is given by ,,' '
II
'-.M"'m=0.7
~
\ \
e_,,,mx , I ), + ,o ~m = 2.0
µ = E(X) = D:f(x) = L x - - X!
0.25
1" /\\'
, :" ' \\ \\ /< \,,.__ - .,,,.m-
111 I - 4.0
m3e_,,, m"e_,,, I ,w ' b-
= 0 + me_,,, + m1 e_,,, + - - + - - +
~ \
l r, ' '
I ' '
or .,JY \_ \ ' ';) ' , ,
- me-m r·1 + m + _m_' + m' : ! ] 3! ...... 0- < 1 0 ~ ►--
I
.... _ -
2! 3! ...... 2 4 6 8 10 12 14
X
= me-m e"' = m The Poisson distribution can frequently be used to approximate the binomial
Thus the mean of the Poisson distribution is m. distribution when II is large and p is very small.
 ~
r
:'!!.. .
BUSINESS STATISTICS
494 PROBA BILITY DISTRIB UTIONS 495 ,
Form of the PoissOD Dlstrlbudon (Iv) Probability of more than three defectives
Like binomial distribution, the variate of the Poisso n distrib ution - P[X :i!: 4) • 1 - [P(X- 0) + P(X • I)+ P(X- 2) + P(X-
is 3))
also discret e one, i.e., it takes only integra
1, 2, ....... occurrences are given by the successive terms of the expans o
l values . The probab ilities of ' • ml
- 1- [ r + fM-- + e-- +
ml
e-
]
ion. 2 6
e -m (1 + m + m 2 + m 3 + ......... + m r + ••••••• ·)
2! 3! r!
... 1- e_,,, [ 1 + m + ml + ,,.3]
2
= 1 - 0.7788(1 + 0.2S + 0.03125 + 0.0026)
6
- 1 - 0.7788(1 .28385)
This can be wutten in tabular forms as follows : = 1 - 0.99986 = 0.00014
Probabi lity Illustration 8. A factory produces blades in packets of 1O. 1be probabil
Probabil ity No. of occurrences ity of • blade to be
No. of occu~nc es p(x} defective is 0.2%. Find the number of packets having two defective blades
p(x) (x) in a consignm ent of 10,000
(x) packets.
0 e• 4 -
m 4 e-m
4 !
Solutio n. m = np ,.. 10 x .002 = 0.02
P[X = 2 defective blades]=
e~·02 co02)2
•
1 mr 2
m2e_,,, = 0.9SOl x0.0()4 = .49()1 X .()()()4 = 0.000196
2xl
2 2! Therefor e, the total number of packets having two defective blades
in a consignm ent of
mre-m I 0,000 packets is
3 -
m3 e-m
3!
r - r'
u Illustration
10,000
9. What
X 0.000)96
probabili ty
= 1.96
model is
= 2
appropri ate to describe a situation where 100 misprint s
arc distribute d randomly througho ut the 100 pages of a book ? For this model. what is the probabil ity
that a page observed at random will contain at least three misprints
where e = 2.7183 and mis a constan t called the parame ter of the djstrib ?
ution. (MFC. De/hi Univ.. /989)
m = the averag e numbe r of occurr ences of an event.
Solution . Since there arc 100 misprint s in 100 pages. it implies
that there is only one
mistake on the average in a page. Therefore. the probability of being
The above table gives probabilities of 0, 1, 2, occurre nces respect ively. If a page contains large number of words and n the number of words a misprint is very small u
in 100 pages will be very large.
we want to know the expected numbe r of occurrences, we have to So in this case 'probabil ity of being a misprint is small and n is
multip ly distribution is best suited here.
very large. therefore . Poisson
each term by N. i.e., the total number of observations. Average or expected number of misprints in one page is
Illustrat ion 7. On the average, one in 400 itmes is defective . If the items are packed in m = np = 100 x .01 = 1
boxes of 100, what is the probability, that any given box of items
will contain : e.... = e-• = 0.3679
(i) no defective Probabil ity of at least three misprints in a page is
(ii) less than two defectives
= P[X ~ 3) = 1 - P<,X < 3) = 1 - [P(X = 0) + P<,X = 1) + P(X = 2))
(iii) one or more defectives
-I •
(MBA. Delhi Univ.. 2000)
(iv) more than three defective s
= 1- [ e-• + e-• + ~ ! ]
Solutoln. Here p = 1 , probability of a defective item which is = 1 - e-• (2.5) = 1 - 0.3679(2.S)
400 very low.
I - 0.9198 = 0.0802 .
:ic
n = 100, number of items packed in the box which is quite large
Fittin g a Poisso n Distri bution . The proces s of fitting a Poisso
distribution is simple. We have just to obtain n
100 the values
m = np = = 0.25 ; average num~er of defectives in a box of 100 items of m and calcula te the
400 probability of zero occurrences. The other probabilities can be easily
calcula ted
(i) Probabil ity of no defective by the recurre nce relatio n as follows :
= P(,X = 0) = e__,,, = e--0·25 = 0.7788 (From the table given in the appendix ] -m X
e m
(ii) Probabil ity of less than two defectives /{x) =. x.'
= P[X s 1] = P(X = OJ + P(X = 1]
-m x+l
= e-m + me_,,, = e_,. () + m) e m
= 0.7788 (1 + 0.25) = 0.9735 f(x+l) = (x + 1) !
(iii) Probabil ity of one or more defective s f(x+ 1) m m
= P[X~ 1] = I - P(X= OJ= I - e-m = I- 0.7788 = 0.2212 f(x) = x+l or/{x + 1) = x+l f(x)
•.l
 496
when X = 0, /{1) = m/{O)
m m2

BUSINESS STATISTlCS
PROBABILITY DISTRIBUTIONS
implies, describes the probabilities of the
num ber of trials likely to be required
~
X = 1, /{2) = --,:f (l) = T/{ O) in order to obtain a fixe d number of succ
when To derive the probability mass functionesse s.
of the negative binomial distribution,
m m3 we proceed as follows : suppose that x
X = 2,
/{3) = -/{2 ) = -/{ O) trials are required to obtain exactly k
when 3 6 successes (x ~ k ; i.e., x = k, k + l,' k
+ 2, .....). Then, clearly, the xth trial
and so should yield the kth success, the previous
[Link] relation
This (:c- 1) trials should give the rem aini
ng
provides a very convenient method for fittin (k - 1) successes and (x - 1) - (k - 1)
Poisson distribution. The only probability we g the = (x - k) failures in som e sequence of
need to know is/(0) = e-"', where
m is the only parameter of the Poisson distr successes and failures. Clearly, this can happ x-1 )
Multiplying by N (the total ibut ion. en in ( k- l distinct ways. Now
frequency) each probability of the Poisson obsei;ve that each particular sequence has
distribution, we get the expected frequencies k successes and (x - k) failures in
Ulastration 10. The following table gives
for respective probabilities. itself and hence its probability of occurren
the numb er of days in a 50-da
ce is fl ( 1 - p y...t. As the sequence of
whkh ,uwmoMc ,cc;dcnt> occun<d ;n a y perio d durin trials are Bemoullian, the probability that
c;ty. flt Po;s,on distribution to the data , exactly x trials will be required to
No. of accidents O
g obtain k successes becomes
1 2 3
No. of days 4
Saltl on.

2 1 18
(MBA 7
. Sukhadia, /992 ; MBA3. Kumaon Univ
FITTING OF POISSON DISTRIBUTION
1
•• 2000) • (X - 1)
k _ l P l ( 1- P)x-l ; x = k, k + 1. ...
Thus, if x is the random variable ·corresp
X f fX onding to the number of trials
requ ired for obs ervi ng exactly k succ
esses in an indefinite sequ enc e of
0 21 0 Bemoullian trials, its distribution is said
I 18 18 to be negative binomial and has the
following probability mass function.
2 7 14
3 3 9
4 1 4 X- f'\J
P[x ; k, p] = ( k _ l p t ( 1- p ).r-t
if x = k, k + 1, ..... .
N=5 0 "i.f X= 45
= 0, othe rwis e
_ "[Link] 45 where 0 < p < 1 and k ~ l is a fixed inte
m = X =N = ger.
= 0.9 The negative binomial dist ribu
50 tion arise s in practice where observation of
/(0) = e- = ci-9 = 0.4066 successes take s place as a waiting-time
/(1) = m /(0) = (0.9) (0.4066) phenomenon.
= 0.3659 For the negative binomial variate, it is easy
m 0.9 to prove that :
/(2) = /(I) =2 (0.3659) = 0.1647 k
2 Mea n = - and variance =
k(l -p)
m 0.9
/(3) =3 /(2) = 3 (0.1647) == 0.0494
p p
2
Illus trati on 1 t. A market resea rch agency that
found, from past experience, that there conducts inter view s by telep hone has
0.9 is a 0.40 probability that a call made betw
0.9 5.30 P.M. will be answered. Assuming een 2.30 and
/(4) = 4 /(3) = 4 (0.0494) = 0.0111 a Bcmo
(a) Wha t is the probability that an inter
ullian proce ss :
viewer's tenth answer comes on his twen
In order to fit Poisson distribution, we shall (b) Wha t is the expected number o'f tieth call?
miltiply each probability by N, i.e.,SO calls neces sary to obtai n seven answ ers '1
Hence the expected frequencies are : . (c) Wha t is the probability that an interviewer will receive his first answer on hts third
X: Solu tlon. Let success denote an answer call?
0 1 2 to a call. Then p = 0.40.
f: 0.4066 X 50
3 4
0.3659 X 50 0.1647 X 50 0.0494
X 50 0.011 I x SO .x-1 )
= 20.33 = 18.30 (a) Prob. (10th answer comes on 20th call)
= 8.23 = 2.47 = 0.56 = (k_
1 l (I - pY - •
Negative Binomial Distribution where x = 20 and k = 10
Where as the bion omi al dist ribu tion desc
ribe s the prob abil ities of the = 20-1 ) 10 10
num ber of succ esse s like ly to app ear ( 10-1 (0.4) (0.6) = 0.05856
in a sequ enc e offued num ber of
Ber nou lli trial s, the neg ativ e bino mia
l dist ribu tion , as the nam e itse lf (b) Expected number of calls for seven k
answers = -p
498
BUSINESS STATISTICS
PROBABILITY DISTRIBUTIONS

,,
499
k-1 aodp-0•4
where (I/) What It the probability that all of them will be accepted 'P
_2-._~-11- ,
.0·4 Solatlon. (I) Here n-6, .r,-2, x.-2;and x 1 -2
4
Tht' rcqubcd probability 11 aiveo by
(c) Prob. (ltt .aoawer OD 3rd call>-( !:! ) 1
(0-4) (0•6r-o·1440 61
- 21 l f 2 i (0·04)'(0·06)'(Q·09,.
MaJtiaomial Distribution -90(0 00000,76)(•8 J)-0·0004199
The multinomial distribution is the multivariate analoguo of the (Ill) Herc ,r-6, .ir1 -0, x,-o. and x 1 -6
binomial distribution. !t !5 o_ne of the ~mpl~~ ~ut. ~st ~portant . 6
discrete multivariate d1strtbut1ons. The binomial distribution anses from Prob. (all will ~ acceptable]- 0 I 0 I I (0·04)•(0 06),0-90>--·.531 4
a random experiment in which a finite sequence of n repeated and inde~ 6
pendent Bernoulli trials are conduc~, each trial resulting in only one or HypergeoD1e tric Distribution
two possible outcomes, success or failure. The multinomial distribution
arises when the number of possible outcomes of a single trial is generalised The Hypergeometric distribution arises when a simple random
1Jample without replacement ·is drawn from a dichotomou.s pop~lation
from 2 to k. (i.e., one whose elements can be divided into two mutwdly ~elusive
Let Eu Es, ...... ,E-. denote the k possible outcomes in a single trial categories) consisting of finite number of elements in each of the two
of the experiment. Let n repeated trials be conducted and the various categories and the random variable observed is the number of items of one
outcomes noted. Let X, denote the number of times the outcome E ia particular category preseqt in the sa~ple. Suppose a box contains numben
observed with correspondin' probabilities p, (i=l~ 2, ••• , k). As 'the of a manufactured product, n1 of wliich are good and n-n1 are defective.
outcomes are mutually exclusive, Suppose a sample of size R is drawn, then,the number X of good items in
X1+X1+•···••+X1t;=n . th~ sample is a random variable taking values 0, 1, 2 ...... , K with
and P1+P1+····••+P1t;=l probabilities, • •
~en th.e jo~t pro~ability mass function of X X1 Xi is call d
i )( ~~i)
1 , , •••••• ,
the multtnoIDial distnbution and is given by e P(X)= (
-
if X=O, 1, ... ...K.
P[X. 1
=x1, X1 =xs,····••; X1t;=X1t;]- X1 ! X1n !!... x1; r PiX1 p I.x, ··•PkX1 { n )
\K
where Xi, Xs,·····•,Xk are non-negative integers such that =0, otherwise.
k This probability mas, function is called hypergeometric distribution.
:t X,=n
i=l •Ii is easy to •verify. that the mean and the variance of the above
fflllStratloa 12. Consider the production of b 11 b • distribution are given by :
diameter or. which should be 0·2500 inch. Because ~f [Link] of a cc_rta!~ tl'c>;e, the
~anufactur,ID.i process, the beariogs arc classified t~: '?:c~eot. varia~1hty •JD the Mean= ni and Varianc-e=K!!!. 11 -ni· n-K
acceptable ,r they mea,ure leH than 0·2495 inch more th er~~~ ~vers1ze' and n . n a -
n-1
twe~ 0·2~95 and 0·2505 inch, respectively. Suppose the produ~io· tDch, and be- Dlustratlea 14. Certain missile components are shipped in Iota of 12. Three
nnp 11 auch that 4C¾ of the bca • d • n process for three
bea
are aec:eptable. If 100 of these bearin ::1r!s~~t:" ersize. 6% _are o~ersize, and 90%
of 1ett1og x 1 uodersizc, x oversize, a:d Xa a~pta:i:\,~~~~f :i:!\what ... -the probability
components arc selected from each lot and a particular lot ls accepted if nooe of
the~threJ components ,elected is defective.
1 (a) What. ·i• the probability :that a lot wiU:-be accepted· if it contain• 5
Solatioa. ~ere n• 100, Pi-0·04, P,-0•06 and •0·90 defectivcs ? '
The required pr~babihty is given by • Pa (b) Wh,t ls the probability that a lot will be rejected if it contain, g
defectives ? •
P(xh x,, x,)-• .x~ l 100 I
Xa I X1 l (O•o4{1 (0·06)x. (0·09).:r. (c) Let Xbe a random variable deootin1 the number of defective• lo a ample
of 3 C\)mponeota selected randomly from one of the above lots. If the lot coota1oa
whert. 0<:x,<;100; l•l 2 3 4 defectives, specify the probability fuoc:tion .It.~).. Presc,nt the probability distribu-
100 O I
tion (i) as a mathematical e~pres1l011, and (II) in tbe form of a table. .
and .Z x,-100. (d) Uqder the conditi0DI stated in (c) abo"•· what la the upected numbor of
1-1 defectives in a 11mple of 3 componoat1 T
Dlustratloa 13. In the above illustration, suppose 6 [Link] sampled fri,m Solution. (a) Here n-12. 111 -s .
the process :
•The lot will be accepted if tlie 3 com-.,ooeate ~aodomly 1clected from thla lot
(i) What i1 the probability that there will be two bearinra of eacb catcaory T laas no de(ectivee~. Tbo required proita~Uity i1, Lbacl\lre, •
 (fl
500
BUSINESS StATISTICS PROBABILITY DISTRIBUTIONS c-9 9 4 r: ....J ·~--
who first made reference to it in 1809. Throughout the 18th and 19th ce~tnrie~!.
501
~~
·-J
_,;11·
( ! )( ! )_2- -0·1591
,·
various efforts were made to establish the normal model as the underlying law
ruling all continuous rahdom variables--thus , the name Normal. These efforts
( 1:) 44 failed because of the false premises. The normal model has, nevertheless,
(b) Herc 11-12. 111-9 become the most important probability model in statistical analysis.
Tbe Jot will be rejected if at leaat one of the c:c,mponenta randomly cboaea The normal distribution is approximation to binomial distribution. Whether
from tbl• lot is defective. or not p is equal to q, the binomial distribution tends to the form of the continuous
Thia probability is given by curve when n becomes large at least for the material part of the range. As a
!/~ ~
1-Prob [non• it defective]
matter of fact, that correspondence between binomial and the normal curve is
surprisingly close even for low values of n provided p and q are fairly near
-1- ( >-1- 2ii -1-0·0045-6·!) 95' equality. The limiting frequency curve, obtained as n, becomes large and is
called the normal frequency curve or simply the normal curve.
(c) Tb~ lot contains 4 defectivea. therefore.- A random variable Xis said to have a normal distirbution with parameters
Number of defectlvca-4 m (mean) and o 2 (variance) if the density function is given by:
Number or non~dectivca- 8 _!f X-µ )l
x-random variable denotioi the number of defective• in a aampte of s 1 -2\
componenta from the above lot. y = f (x) = ..Ji; e O
- oo < X < oo
(J 21t
The required probability function 11
where y = the computed height of an ordiante at a distance of X from the
'( ~ )( 3~X) mean.
/(x)- ( ~2 ) ; X-=0.- 1. 2, ~
a = Standard deviation of the given normal distirbution.
-=O. otherwise 1t= the constant= 3.1416; ..Ji;= 2.5066.
The required table of probabilities is
e = th~ constant = 2.7183 (the base of the system of natural
X /(,c) logarithm).
µ = Mean of the given normal distribution.
In symbols, it can be expressed as
0 0·254S
1 0·5091 ' X~N(µ, o)
2 0·2182
3 0-0182 This is read as : The random variable X follows normal distribution with
mean µ and standard devation o.
If we draw the graph of normal distribution, the curve obtained will be
/ Z/t~)-1
known as normal curve and is given below.
(d} Tac expected number or defective, lo , sample of J component• 1/(x)
•K.~
.. ,.
where K-3. r.1 -4 and 11-12
• 3x4
-12"-1.
Normal Distribution ·
The Normal* distribution was discmered by De Moivre as the
limiting case of Binomial model in 1733.. It was also known to 4place no
later than 1774, but th~ougb_ a historical error it _has been credited to Gausa µ X
The graph of y =f(x) is a famous 'bell shaped' curve. The top of the bell is
. -The word normal docs not simply mean that the other distributioos'ar• directly above the meanµ. For large values of o, the curve tends to flatten out
abnormal. It is timply the na1ne customarily sivcn to this diatrib~ion. Somctimu
the name Gaussian ia uacd io1tcad of nol'mal tQ [Link] con(aeiou.. and for small values of o, it has a sharp peak.
 502 BUSINESS STATISTICS
PROBABILITY DISTRIBUTIONS
503
When we say that curve has unit area we mean d. that the total frequency N
. To obtain ordinates for particular 1stn·button• th d.
1s_ equated to l. "' e or 1nate
plied by N. The equation to a nonnal curves or ~ will follow the normal distribution with mean
given by the above ,onnu1a are multi
. . . . . b npq zero and variance
corresponding to a particular d1stnbut1o n is given y . one.
- N e-¥X~tir Similarly, Poisson distribution also approaches a norm
al distribution with
y- cs..f2n standardised nonnal variable, i.e.,
z=
X-m
N
___,,, ,,~~ N(O, 1)
~ in the above fonnula is equal to the maxu
Thequantity csv21 • num d. ✓
m
Of mate
•
t
In other words, ~m will follow the normal distributio
0 f the normal cwve which will always occur at the mean of the n with mean
distribution.
A random variable with any mean and standard devia tion zero and variance one.
can ~e ~fo rme d
to a [Link] normal variable by subtracting the mean and d1v1ding by the The Standard Deviation and the Normal Curve
standard deviation. For a normal distribution with mean
µ and standard deviation In any normal curve, an exact percentage of obsei
cs, the standardized variable z is obtained as vatio ns in the distribution
falls within ranges esta~lished by the standard devia tion in conju nction with the
X-µ mean. If we cut through the distribution at one, two, and
z= -- three standard deviation
cs away from mean, on both sides, we obtain the areas of
the distribution which
A value z represents the distance, expressed contain certain percentages of all the obsetvations.
as a multi ple of the stan ~d
deviation, that the value X lies away from the mean. The In a normal curve, between the range of arithm etic mean
standardised variable plus 1 standard
z is called a standard normal variate which has mean zero deviation and minus 1 standard deviation, i.e.,µ ±
and standard deviation 1 cs cover s 68.27% of the
one. In symbols, if observations in the distribution and 34.13% of obser vatio ns will fall on. either
side of mean. Similarly, µ ± 2cs covers 95.45% of obser
X ~ N (µ, cs) vations and µ ± 3cs
covers 99.73% of obsetvations.
then z ~ N (0, 1) The table shows the area of the nonnal curve betw een
The probability density function of the standard nonn mean ordinate and
al variate z is given ordinates at various sigma (cs) distances from the mean
(µ) as percentage of
by . 1 _..!_z2
the total area.
y= f(z)= J2n e 2
-oo< z<o o
AREA RELATIONSHIP
Distance from the Perce ntage of
Relation between Binomial, Poisson and Normal Distr mean ordinate total area
ibution
The three distributions, namely Binomial, Poisson and 0.5 cs
Nonnal, are very 19.146
closely related to each other. As explained earlier when
n is large and the 1.0 (J'
34.13 5
probability p of occurrence of an event is close to zero
so that np remains a 1.5 cs 43.31 9
finite constant, then the binomial distribution tends to
Poisson distribution. 1.96 cs 47.500
Similarly, there is a relation between binomial and norm
al distributions. 2.00 cs 47.72 5
Nonnal distnbution is a limiting fonn ofbinomial distributio
conditions : n under the following 2.5 cs 49.379
(1) n, the number of trials is very large, i.e., n ➔ «>,an 2.5758 cs 49.50 0
d 3.0 cs
(2) neither p nor q is very small. 49.865
In fact it can Thus the two ordinates at distance 1.96 cs from the mean
be prove d that the binomial distribution approaches a normal on either side
distribution with stand would enclose 47.5 + 47.5 =95% of the total area and
[Link] nonnal variable, i.e., two ordinates at 2.5758
a distance from the mean on either side would
X-n p
the total area. The area enclosed between
enclo se 49.5 + 49.5 = 99% of··
z= -== -~N (0,l ) ordin ates at 3cs distance from the
✓npq mean on either side would be 49.865 + 49.865 = 99.73
% of the total area.
 BU SIN ES S STATIS TIC
-
50 4 S
PR OB AB ILI TY DIS TR
M om en t• of No al Di str ib ut ion IBU TIO NS
tb e ~ . . ll od d order moments 505
Fo r the normal d1stn ab ou t me an ar c 7. First and third quartile
zero and given by the rel ~uuon, a s are equidistant from the
median.
ati_onu.
1'1 -r• = ru 6 =•••. •• ••• •• •= l't *l =0 8. M ea n de via tio n ab ou t me
t bo ut mean is given an is= i a or, more precisely, 0.797
an d the eve n ord er mo roe nsa by the r elau•on of standard deviation. • 5 9 times
~, .= 1.3.S ...••• ••• (211.;,_ l)a '" 9. Li ne ar co mb ina tio n of independent normal va
Io particular, we ha ve variate, i.e., if X, and X riates is also a normal
l'-t = l •a '= a' ; I', = 1 3 c 3 '. are tw o ind ep en de nt normal variates an d a, an
Therefore, moment coeff • •a = a • given constants, the 2
icient of skewness ~1 is n the lin ea r co mb d a2 are
normal distribution. ina tion a~ . + a, 7 will als
~ _,.,.,~ ==0 given_ by 2 o follow a
t"l- I • [ ·: l'a =O ] • 10. All odd moments of
#'I the normal distribution
an d moment co eff ici i.e ., are zero.
en t of ku rto sis is given by µln +I = Q
3a ' 11. 13, = 0 and 13 (n = 0, 1, 2, ...)
~. =~ = a' =3 2 =3
Hence ~ =0 sho ws th at .. since 13, = 0, therefore,
an d ~ =3 indica no rm al . dis tri bu tio ~ }S perfectly sy mm etn the no rm al distrib uti on
1
tes the normal curve 1s ca l = 3 im pli es tha t no is perfectly symmetrical an
curve 1is platylrurtic an d me sok urt1c rm al cu rve is neither leptokurtic no d 132
if ~1 ~ 3, the curve is lep 10 s~ape.- If '32 < 3, th e 12. M ea n± cr, me
r platykurtic.
tokurt1c. an ± 2cr , an d me an ± 3cr covers 68.27%, 95
Pr op er tie • of th e No 99.73% area respective .45%, an d
rm al Di str ib ut io n ly.
The following are th Im rtance of Normal
po
e im po rta nt pr op erties of th e no rm al cu Distribution
the normal disUibution rve an d Th e no rm al dis
: • tri bu tio n ha s great significance in sta
1. The normal cu rve is of the following reasons tistical wo rk because
curve is o s mm elr tca l ab ou t th e me an skewness :
a ong us ver 1ca
of cases below the me run s, t e tw =0)~ If th e 1. Th e normal dis
cases above an in a normal dis triobua ves wt co1nci e. Th e nu mb er called central limit the
tri bu tio n ha s the rem arkable property stated
the me an , tio n is eq ua l to th e ore m* , wh in the so-
of the curve for a po which makes th e me an an d me dia n coincide nu mo er of
sit ive . Th of wh ich is the sa mp
ich ass ert s that certain statistics, mo
st important
de via tio n of 3 un its is th e sa me as th e he e he igh t le me an an d sample variance, ten ds
curve for a negative de via igh t of th dis buted as the sample
tri to be no rm all y
tig ,n of 3 u_n its. _ e size becomes large.
2. Th e height of the _ 2. Ev en if a va ria ble
Hence the no rm al cu rve is at its ma is no t no rm all y distributed, it can sometim
me an an d mo de xim um at th e to normal form es be brought
normal distribution mean of th e no rm al dis tri bu tio n concide. Th us me an .
fo r a of Xi s sk
by sim ple tra nsf orm ati on of variable. For example
, median an d mo de ar e ew ed , the dis tri bu tio , if dsitribution
3. There is one all eq ua l. n of .J x might come out
to be normal.
the mean. Th e heighmaximum po int of th e no rm al cu rv e wh ich 3. M an y of the sam pli ng dis trib uti on s like Student's t, F, etc. als
from the mean. To e t of the cu rv e declines as we go in eit he occurs at normal distribution. o tend to
it neyer to~ches it. curve ap pr oa ch es ne are r an d ne are r to th r direction 4. Th e sa mp lin g the ory
o~ e1~her ducction. In oth~r words, th e cu rv e is as ym pto tic e ba se bu t assumption tha t sam
an d tes ts of hypothesis are ba se
d up on the
dir ec tio ns.
He nc e its ran ge is unlimited __or infiniteto th e ba se µ an d variance a 2•
ple s ha ve be en dra wn from a normal populat
on bo th ion wi th me an
~- Si~ce the re is on ly 5. No rm al dis tri bu
curve um on e ma xim um po int tio n fin d large applications in Sta
1s mo da l, i.e ., it ha s only on e mo de . the ref or e the no rm al 6. As n be co me tistical Quality Control.
S. Th e points of inflectio ,. • s lar ge , the no rm al dis tri bu tio n se rv es as
n occur at ' approximation for a go od
ma ny dis cre te dis trib utions (such as Binomial,
X= J£ ±a , )' J'( x) = 7. In theore tic al sta tis tic Poisson, etc.).
1 e- t s ma ny problems can be solved
ay 2n assumption of a no rm al only under the
. b6. _As . distinguished _p op ula tion. In applied work we often
from binomial an d Po develop ed un de r the no rm find that methods
!a na le ~s dis iss on al pro bability law yield satisfact
IS
cre te,
a continuous one.
the va ria ble dis tributed ac co rd ing to dis tri bu tio ns where the assumption of a no rm al po pu lat
ory results, even when_
th e no rm al cu rv e pr ob lem ca n have ion is no t fully met, despite the fac
a for ma l so lut ion on t that the
ly if such a premise is hypo
thesized.
• Sec Chapter on Sam
pling and Sampling Dis
tributions.
506 BUSINESS STATISTICS
PROBABILITY DISTRIBUTIONS
507
8. The nonnal distribution has numerou~ mathematical properties which
X-µ -
make it popular and comparatively easy to ~~pulate. For example, the moments Solution. z, = - -- IS~=- 0.S
9 C1 10
of the normal distribution are expressed m simple form. The nonnal curve i
reasonably close to many distributions of the humped type. If, therefore w 40-20
are ignorant of the exact nature of a humped distribution, or know the bu~ roni: z2 • ----+
10 - 20
•.
find it mathematically intractable, we may assume as a first approximation that Consulting the table we find the areas corresponding to the z's arc 0.191 S and 0.4772 and
the distribution is nonnal and see where this assumption leads us. thus the desired area between X1 a IS and X2 • 40 is (0.1915 + 0.4772) • 0.6687 u given in the
figure.
The admiration of nonnal distribution has been beautifully expressed by
Illustration 18. How ma~y workers have a salary between 400 and 650, if the arithmetic
well-known statistician W.J. Youden in the following words : a mean is 500 and standard deviation is 100, and number of worker is IS,000 if the salary of the
worker is assumed to follow the nonnal law ?
~ -
THE Solution. -~""'"'"':'-
NORMAL
LAW OF ERROR
STANDS OUT IN
THE EXPERIENCE OF
MAN~NDAS ONEOFTHE
BROADEST GENERALISATION OF
NATURAL PHILOSOPHY. IT SERVES AS THE
GUIDING INSTRUMENT RESEARCHES IN THE
PHYSICAL AND SOCIAL SCIENCES AND IN MEDICINE, ,'!
AGRICULTURE AND ENGINEERING. IT IS AN INDISPENSAB LE
TOOL FOR THE ANALYSIS AND THE INTERPRETATION OF
THE BASIC DATA OBTAINED BY OBSERVATION AND EXPERIMENT
Artistically enough, it gives us the shape of the nom1al curve also.
zI =
-
400-soo
100
=- I (let\ of the mean)
Area under the Normal Curve ~-~ .
z2 = -:--:-:-
100 = 1.S (nght of the mean)
Since for different values of µ and CJ, we shall have different nonnal
curves, therefore, it may be very difficult to find areas under the normal curves of workers fall between 400 and 500 and 43.32% fall
From
between 500the
and [Link] find that 34.13%
table
for different pair of values ofµ and CJ. Hence the areas for a normal curve are :. 34.13 + 43.32 • 17.45% of workers have a salary between 400 and 650.
tabulated in terms ofthe standardized normal variate z. As any normally distributed :. Number of workers getting a salary between 400 and 650 is given by
random variate X with parameterµ and CJ can be transformed to the standardized .1145 X 15,0()(},. J),6)8
normally distribut~ random variate z, therefore, the table given in the appendiJ Illustration 19. A large flashlight is powered by S batteries. Suppose that the life of a
battery is normally distributed with µ • 150 hours and a=- IS hours. The flashlight will cease
under the heading of area under the normal curve may be used. functioning if one or more of its batteries go dead. Assuming the lives of batteries arc independent.
This table in the appendix contains the probabilities for the area under W what is the probability that flashlight will operate more than 130 hours ?
normal curve between mean z = 0 and any other specified value of z. As W Solution.
130-150
z ... -X-µ = --:-:-- = - 1.33
normal curve is symmetrical, therefore, the area under the normal curve it • a IS
given only for half of the positive side of the curve. From the table we find the 40.82% of u1c batteries will operate between 130 and 1SO hours.
For example, corresponding to z = 1, the area under the normal curve b
~ven as 0.3413 (from the table in the appendix). Therefore, for z =-1, the area
IS also 0.3413.
Hence Pr [ -1 ~ z ~ + 1] = 0.3413 + 0.3413 = 0.6826.
~e following are some of the examples to illustrate how tables are to be
used m orde~ to obtain area under the nonnal curve. . .
Illustration IS. Find out the area under the normal curve for z = 1.54.
. SolutJon: If we look to the table, the entry corresponding to z = 1 S4 is 0.4382 and tJiil
gives the area m the shaded region in following figure between z = o and ; = 1.54. 180
 BUSINESS STATISTICS
508 PROBABILITY DISTRIBUTIONS
p, [Fla,hHght Hf•> 130 h,s,J = p, {[Link] batte,Y 509
lffe > 130 h,s.J
Since the life of batteries arc independent 5 Solution. CALCULATION OP Z AND o
,., Pr [each 90821
. . batter)
· ' life>
fl 130 hrs.)5
ht" ht will"' operat
[0- e more than I 30 • given
• X
hours ,s by I d fd
Hence the probab1ht)' that the as ,g
(0.9082)5 •
61
62
2
10 _,
-6 - 12 '"'
72
-,0 2'0
Applications of the Norm 63 11 -4 -44
al Distr 176
The nonnal distribution is mostibuti on
ly used for the follo
• • 64 38 -3 -114 342
· t or "fiit'' a distribution of measuremwing purp oses . • 65 S1 -2 -114
ent unde r certam 66
228
1. To approx1ma e 93 -1 - 93 9)
cond2. 67 106 0
itions.
To appro ximate the binomial distribution and other d",screte or 0 0
• uous
contm 68 126 +1 +126 126
probability distribution unde suita 69 109 +2 +218 436
3. To approximate the rdistri ble cond itions.
butions of means and certa•in othe r statis • • 70 87 +3 +261 783
tic 71 15 +4 +300
calculated from samples, especially large samples. 1,200
72 23 +5 +115 515
73 9 +6 + 54 324
Fittin g of Norm al Distr ibuti on 1 74 4 +1 + 28 • 196
In order to fit a normal distribution to the given data we first calculate the N-15 0
mean µ and stmdard deviation a from the given data. J/d=-615 J/d'- 4801
Then the nonn al curve
Jfd 675
fitted to the given data is given by ~-A +,r -61+
750 -67·9
y= 1 e -~X~ µr
-,-;::-:: -~< X<o o 0 r'1,fd'
-,\J - i r -
('/~td )'
N r 480-
-"I 750 --,-(1so
6~15~).
a-v21 t
To calculate the expected nonnal frequency -.[o ·4-· st -~57 59-2 •3
, we first find the standard
normal variates corresponding to the lower units of Z±lo -67•9 ±2·3 -(66· 6 and 70•2)
each class intervals. Then ~umb er of traine es havio1 statu re io this 'ta-~
the areas under the nonnal curve are computed from
the tables for each standard -93+ 106+ 126+ 109+ 87-52 1
normal variate. Finally, the areas for the successive
class intervals are obtained
by subtracting the successive areas and multiplyi .·.Pro porti on-~ ~~ •0·69 -69%
ng these areas by N, we get
the normal frequ encie s. • Z±2o -67·9 ±4·6 -(63· 3 and 72·5)
The following illustration shall illus trate the Numb er of traine es havin g statur e in this raoae
appl icati ons of normal
distribution : - t t+J8 +57+ 93+l 06+1 26+ 109 +87+ 75+2
3-72 5.
lllustratlon 20. The following table gives the
distrib ution of heigh t statur e among the ••• Propo rt~on - ;~~ -0·96 ~96%
management trainees in Delhi :
X±3o -67•9 ±6·9 -(61 anJ ~8).
Height No. of Heigh t No. of , Numb er of traine es havin
(in Inches) traine es g statu re in this ran~ 7SO and hence . 100% . (n a
(in inches) trainees norm al di1tribution the propo rtion lying betwe
and 99% rcsP" tivcly . Henc e the given distri en tbe~l imit1 is abou t 68%, 95%,
61 2 68 126 butio n is appro xima tely norm al.
Unif orm Dist ribu tion
62 10 69 109 . .
63 1l 70 87 A conti nuou s valued rand om varia ble
64 38 71 75 distr ibuti on with param eters a and b if its Xis said to have a unifo rm
prob abili ty dens ity func tion is
given by
65 57 72 23
66 93 73 9
67
P(x )=- bl, if a<x <b
106 74
-~ -a
4 --
Test the normality of the distribution by compa
ring the proportion of cases lying between =0, otherwise
X ± la, X ± 2a, X ± 3a for the above distrib
ution.
 BUSINESS STAT1STJcs
510 PROBABILITY DISTRIBUTIONS 511
The uniform distribution is also know n as the 'Cons tant
Distribution', as_ Probabil' • We note that the only condition on A is that it shoul
the proba bility . is. const ant [equa l to 1/(b- a)] at ev:ty d be a posit ive real
point or tbe conunuum (a, b) and 1s independent of whate numb er. Henc e by giving different value s for
variable may take within the interval. ·Yet anoth ver value t cy A, differ ent expon entia l dis- -
er name for the un;r0 he tribut ions can be specified. This would help us
distribution is 'rectangular distrib ution' , for the 0 to under stand the natur e of
shape of the
probability density function is rectangular, as shown below curve £ ;: the expon ential distribution better. The follow ing table
gives the ordin ates
: • of the expon ential probability density function for x=O
to 6 for A=0. 2, 0.5,
P(X) 1 and 2.
It,-
A 0.2 0.5 1 2
X
'
b-0 i:..·- - - - - - - - - " I
0 0.200 0.500
0
- a
...
b
........
)()
1 0.164 0.303
1.000
0.368
2.000
0.271
2 0.134 0.184 0.135 0.007
3 0.110 0.112 0.050 0.005
It is easy to check that P(x) defined as above is indeed a
probability 4 0.090 0.067 0.018 0.000
density function.
The mean and variance of a rectan
s 0.073 0.041 0.007 0.000
gular distribution are given by 6 0.060
- I 0.025 0.002 0.000
br+-a . 2
Mean ----"2-- • Varia
' nce= (b-a • • )--
12 It can be prove d that
The mean, varian ce and, in fact, all highe r order mome 1. The exponential density function decreases in the range
rectangular distribution depend nts of the O to oo, the
solely on the value s of a and b, the lower maximum ordinate of the curve occurring at x = 0 (the value
and upper bounds of the range of values the variab being A itself ),
interesting property of the unifo le takes. Another 2. Large r the value of A., steep er is the decline in the ordin
rm distri butio n ate, even for
by sub-intervals [of the main interval (a, b)] is that events described
of equal
small value s of x,
probabilities of occurrence. length have equal
This is a direct conse 3. Smal ler the value of A., flatter does the curve becom e
distribution of the total probability over the entirequenc e;, of the constant
range. to X-axis.
and lies close r
The uniform distrib ution arises in practi ce when ever the The exponential distribution arises in practi
of occurrence of the event under considerati proba bility ce when the rando m varia ble
on is const ant whate ver be the studie d is service time -the time taken to comp lete servic
value of the variab le, i.('., all possib le value s of the e at a fillin g
are assumed equally likely. For instan contin uous variable ~tation, grocery or automobile repair shop. It is also used
ce, comm in reliability theor y
specified points has been considered as a rando m uter travel time between
variable with constant
to mode l the life times of comp onent s subje ct to wear.
e.g.. batte ries.
probabilities over a small range of time. transistors, tubes. bulbs, etc. It has also been used to mode
l the distri butio n
Expo nenti al Distr ibuti on . of lengt h of time betwe en successive rando m even ts-th
e time betw een
. .A ~ontin ~ous rando m variab le x is said . to have an exponential arrival of two customers at a service station or the time betwe
distribution with 'parameter A ~fits probability densit en break down s
y function is given by of a mchine.
The mean and variance of the exponential distribution are
P(x)= Ae-"A x if 0<:x<:oo, A>0 given by
1
M ean = i"; V . 1
•=0, otherwU;e. ananc e = ii"
At the outset we shall t th t h
gular) variab le takes value no e a '1'. ereas th'e uniform. (or rt~tan MISCELLANEOUS ILLUSTRATIONS
takes values over an ·
infinit e ;a over [Link] e range , the expon
density function defined as abo~:·. 1~ ~ easy to verify_that ential variable
Illustration 21. The averag e daily sales of 500 branch offices was Rs. 150 thousa nd
the exponential and the standa rd deviation Rs. 15 thousand. Assuming
the distribution to be norma l, indi-
is m eed ~ probability d~nsity function. cate how many branches have sales between :
(i) Rs. 120 thousand and Rs. 145 thousand.
(ii) Rs. 140 thousand and Rs. 165 thousand.
 BUSINESS STATIS
• Ttcs PROB ABILI TY DISTR IBUTI ONS
513
512 Illustration 23. Five hundre d televis ion sets are inspec ted as they come off the produc -
Soluti on. (i) Standard nonna l variate corres pondin
g to 120 is tion line and the number of defects per set is record
ed below
X-µ 120-15 0 No. of defects (X) 0 J 2 3 4
:z=- = =-2 No. of sets 368
a 25 72 52 7 I
Estimate the average numbe r of defects per set and
expect ed frequencies of 0. I. 2. 3 and
4 defect s assuming Poisson distribution. (MBA. S11k/1ndia Univ.. ; MBA. Delhi Univ.. 1997)
Soluti on. For finding the average numbe r of defect
s. we make the follow ing table :
X I fX
0 368 0
1 72 72
2 52 104
3 7 2I
and corresp onding to 145, the standa rd norma l variate
is 4 . I 4
145-15 0 -5
:z = - -15 - = -15 = -3 •• 3
from the table. we find the areas N = 500 "[Link]= 201
Therefore the desired
corresp onding to the values of z are .4772 and 1293
area betwee n Rs. 120 and Rs. 145 = .4772 - .1293 = ,3479• _ [Link] 201
Hence the expected numbe r of branch es having
sales betwe en Rs. 120 thousand = N = 500 = .402
and Rs. 145 thousand are Thus averag e numbe r of defects per set is m = .402
.3479x 500 = 173.95 = 174 approx .
The expect ed freque ncy of getting O defect
(ii) Standard normal variate corresp onding
to 140 is = NP(0) = 500 " e- 402
140-15 0
z=
15
= -
-2
15
= -67
•
But e-..Wl = .6689
rFrom the table]
~-
NP(0) = 500 x .6689 = 334.45
and corresponding to 165, the standa rd norma l Expec ted freque ncy of getting one defect
variate is
165-15 0
Z= =l NP( I) = [Link]( 0) = .402 x 334.45 = 134.45
15
Expec ted frequency of getting 2 defects
Ill .402
NP(2) =
2 . NP( I) = -2- x 134.45 = 201 x 134.45 = 27 .02
Expec ted frequency of getting 3 defects
Ill .402
NP(3) = 3 . NP(2) = -3- X 27.02 = 0.134 X 27.2
= 3.62
Expec ted frequency of getting 4 defects
140 150 165
m .402
from the table. the areas corres pondin g to the z
values are .2486 NP(4) = 4 . NP(3) = 4 X 3.62 = .1005 " 3.62 = 3.36.
and .3413.
Theref ore, the area is .2486 + .3413 = .5899
Hence the , Illustration 24. In a nom1al distribution 31 % of
expect ed numbe r of branch es having sales betwe en Rs. 140 thousand ani. the items are under 45 and 8% arc over
Rs. 165 thousand are .5899 64. Find the mean and standard deviation of the distrib
x 500 = 294.95 = 295 approx . ution. (,\,/BA. Delhi Univ.. /999)
Illustration 11. Eight coins are thrown simult Soluti on. Let mean be µ and standard deviation
aneous ly. Using binom ial distrib ution sho• er, 31 % or the items are under 45. They
that the probability of obtain ing at least 6 heads are lying to the left of the ordinate at X = 45 is 0.31,
is 0.1445 . and therefore, are lying to the nght of the
Soluti on. If eight coins are thrown simult aneous ly, the probab ility of gemng at tea$ ordina te up to the mean is (0.5 - 0.31) = 0.19. The
value of z corresp onding to this area is 0.5.
six heads will be given by the separa te probab ilities
of getting 6 heads. 7 heads and 8 he~
i.e.. = P(X ~ 6] = P[X = 6) + P[X = 7) + P[X = 8)
n = 8; X = 6, 7, 8; p = -l =q
2 119%1 42%
6
= 'C 6(~) (~)\•c, (~) (~) +•c. (~J (~)'
= 28 (~J + (~J (i)'
8 +
45 ~l
l)M 37
= (- [28 + 8 + 1] = - = 0 1445
2 256 •
 I BUSINESS STATISTICS
PROBABILITY DISTRIBUTION
514 S
S15
Hence 45.:..
CJ µ The stan dard normal varia te corr
z= - == -8.5 ... espo ndin g to 1832 is
8% of 1he ;lCms are above 64- Ther (i')
efore, area 10 the ,;gh l of the ord; 1832 -175 0 82 I
Arca 10 1he ten of lhe onl;nate al X nate at 64 ;, [Link]. = 50 so
= = l •64
• 64 up to mean onl;nate ;, (Q.5--0.08)-0-
42 and the value Area to the right of ordinate at 1.64
is
I
I
of z corresponding to this area is 1.4. 0.50 00 - 0.4495 = 0.05 05 I
64- µ ...(ii ) The num ber of pers ons getti ng abov e Rs. 1832 is I
z== - == t.4 ) 0,000 X 0.0505 = 505 I
Hence CJ This is 5% approx.
From equations (i) and (ii) Probability of getting richest
-µ -t- [Link] == - 4S 100
-µ - 1.40 == - 64 JOO
a== 10
= 1000 = 0.01
• 1.9G == 19 or
µ - 0.5 x 10 == 45 or µ == 50
The mean of the distribution is 50
Illustration 15. The income of a grou and standard devi ation 10.
p of 10,0 00 pers ons was foun d to be normally
d;stribu1td w;th mcon Rs. t 750 p.m.
and stand an! dev;ation Rs- 50. Show
had income exceeding Rs. 1668 and thal of th;s group 95%
only 5% had income exce edin g Rs.
lowest income among the richest 100 1832. Wha t was the
? (MBA. Dell ri Univ•• /997)
Solution. ~tandard normal variate
is
X-µ Z=O
z==-
0
Stan dard norm al variate havmg 0.01
Herc
X== 1668, µ = 1750. = 50
\I area to its right = 2.33
X- 1750
2.33 =
50
X = 2.33 x SO + 1750 = Rs. 1866
appr ox.
Hence the lowest income among
the richest 100 is Rs. 1866.
Illustration 26. A workshop prod
uces 2000 units per day. The aver
is 130 kg with a standard deviation age weig ht of units
of 10 kg. Assuming normal distributi
expe cted to weig h less than 142 on. how man y units arc
kg?
Solution. We are g1ven : (MB A. DU. /996 )
µ = 130. o == 10, N = 2,000. X =
142
166 8- 1750
z == - - - - == --82 == -l.6 X-µ 142- 130
so 4 z
50 =-(J- = 10 = 1.2
Arca to the right of the ordi nate
at - 1.64 is (0.4495 + 0.50 00) =
· The expected num 0.94 95.
ber of parsons getti ng abov e Rs. l 668
:: }0,000 X 0.9495 := 9495
This is about 95% of the total, i.e.,
10,000.
0.0595
Areas between z == 0 and z = 1.2 is 0.3849 •
Probability of units having weight
less than 142 kg
== .5 + 0.3849 = 0.8849
Expected num ber of units weighing
less th~n 142. kg
1750 1832 = 2,000 x .8849 = 1769.8 or I 770.
-:.
 aus tNE SS STATlSTlCS
516 111ust<allon 27. PROBABILITY DISTRIBUTION
. er. S 517
There are 600 business students in
the graduate depa1 rtme nt of a univ
shy. and the probability for any Illustration 29. As a result of
student to need a copy of a paru tests on 20,000 electric fans
university library on anY day is [Link] cu ar text boo k from th it was found that lifetime of the man ufac tured by a com pany ,
. )tow many copies of the book shou ld be kep t in the fans was normally distributed with
univers,ty library so that the prob an aver age life of 2,04 0
~
ability may he greater than hours and standard deviation of
0.90 that non e of the studen e 60 hours. On the basis of the info
needing a copy from the library has of fans that is expected to run rmation estim ate the num ber
to come back disappointed ? (Use for (a) more than 2,15 0 hou rs and
normal approximati (b) less than 1.96 0 hours.
to the binomial probability law.) Solution. (a) X = 2150, µ = 2040, a = 60.
Solution- Let X represent the num
ber of copi es of a textb ook
Mean = µ = •P = 600 x 0.05 = 30
requ ired on any day. on X-µ - 215 0-20 40 _ 110
'' z = --
CJ - =· 1.833
60
✓npq
- 60
s.d. = CJ = = .[6oox.0Sx.95 = ['is s = 53.
Therefore. area between mean and 4
ordinate at X is grea ter than (0.9
con-esp0ndmg 10 th,s area standaid -0.5 ), i.e.
normal variate z = t.28, and we
get • o. ,
X-µ
->z
CJ
X-3 0
~ > 1.28
X - 30 > 6.784 or X > 36.784 196 0 204 0 215 0
or
Area to the righ t of ordi nate at 1.833
= 0.5 - 0.4664 = 0.0336
The num ber of fans that is expe cted to run more than 2,15 0 hou
0.03 36 rs
X 20,0 00 = 672 .
(b) X = 1960, µ = 2040, a
o.5 I o.4 = 60.
196 0-20 40
z = = - 1.333
60
Area to the left of ordinate at
1.333 = 0.5 - 0.4082 = 0.09 18
30 :. The number of fans that is expe
cted to run for less than 1.960
hou rs
Hence, 37 copies of the textbook
are required on any day. = .0918 X 20,000 = 1836.
Illustration 28. 1,000 tube ligh Illu stra tion 30. The results of
ts with a mean life of a particular examination are give
120 day s are inst alle d in a new form . n belo w m a sum mar y
factory, their length of life is norm
ally distributed with standard devi Res ult
many tube lights will expire in less atio n 20 days. (i) How % of candidates
than 90 day s? (ii) If it is decided (i) Passed with distinction
together, what interval should be to replace all the tube lights 10
allowed between replacements if (ii) Passed without distinctio
n
not more than 10 per cent 60
should expire before replacem ent (iii) Failed
?
lt is known that a candidate fails 30
Solutlon. (i) µ = 120, CJ = 20, X in the examination if he
= 90 of 100) while he must obtain at obta ins less than 40 mar ks (out
Standard normal variate is least 75 marks in order to pass with
mean and standard deviation of distinction. Dete rmin e the
the distribution of marks, assumin
90-1 20 Solu tion . We hav e to com pute g this to be norm al.
z= 20 = -l.S the mea n and stan dard dev iatio n from the give n
info rma tion . The following diag
ram will help in und ersta ndin g
Area of the curve at (z = - 1.5) solu tion : the que stio n and find ing its
up to the mean ordinate = 0.43
32
Area to the left of - 1.5 = 0.5 -
0.4332 = 0.0668.
Number of tube lights expected to
expire in less than 90 days
= 0.0668 X 1000 = 66.8 = 67
(ij) The vlaue of standard
normal variate corresponding to s
an area 0.4(0.5 - 0.1) is 1•2
X-1 20
20
= - 1.28
X = 120 - 1.28 x 20 = 120 - 25.6
= 94.4 or 94. 40
Hence the 10 per cent of the tube 75
lights wi 11 have to be repl aced We kno w that 30% students get less
afte r 94 days than 40 marks.
\
 ausn11ESS STATlS TlC
S
PROBABILITY DISTR
IBUTIONS
5\8 d" to 519
Thctefore, from th< t>b Sol uti on. Let the dem
le,'o.2(
val200
ue /o 4 and ('00 0 units) be don
co" ""' etc d by X. Th en
area)on
:::: -ing
o.s l
£(X ) = tX P(X)
40 -µ ... (i)
_. ::: : -0. 52 4
= sx20 ·)
- 60 ) 80
310 + 6> 310
Hence 0 <- )
+ 7> <- 120 ) + 9x - 20 ) ·10 \
/\ISO 10% students gel distinct
ion n,arks, ;.,.. 75 or mo 310 + 8> 310
<-
310 + I0 x -310 /I
re
Th«"fo re, rrom
th< table, • value coresponding to
1
= 31 O [ 100 + 360 + 560 + 960
=
0.4(400/o area) l .28 + 180 + 100]
I
75 -µ = 310 = 7.29032
Hence
- (J =
[22 60] = 226
31
1.28
Solving equations (i) and Therefore the expected
(ii), we get demand =- 7 .29032 x
µ:::: S0.17 and [Link] 19. I000 = 729 0.3 2.
4 Illustration 33. In a tow
Hence th< mean of 111< dist n 10 accidents took place
ribution is S0.11 and stan number of accide in a span of 50 days.
dard deviation 19, nts per day follow s Poisson distribution, find Ass um ing that the
mastradon 31, A complex 4, the pro bab ilit y that the
television component has three or more accidents
1<nown ro produce on ave I ,000 joints by ma chi ne in a day. re will be
rage one def ect in fort which is
soldering con Y• The components arc exa min Solution. Th e average
ectcd by hand. If com ed, and faulty number of accidents per
ponents requiring more • day is
whll proportion than 35 com:ctions arc 10
of the com pon ent s will be thrown awaY' (Ap dis carded
ply normal dis trib utio m= = 0.2
Solud• ~· Since the pro bab n.) 50
ility of occurrence of the def
more appropnatc to use nor ect is ver y sm all, therefor<:, it is Prob (3 or more accide
mal distribution as a lim nts) = 1 - Prob (2 or less
iting case of Poisson dis accidents)
Here
m :::: average number of def
tribution. = l- [Prob (0 accident)+Prob (1 acc ide nt) + Pro
ects = np b (2 accidents.)]
=l - [ e • + me •+ ,nl
:::: \00 0 X -
40 ' :::: 25
2 f!•
]
µ= J;= J2 5= 5 =l -e •[ 1+ m +~
]
We know that x-m
Since it is required to find
z = J; ~
N (0, 1). = 1- (!"'-': [ 1 + 0.2 + -0.0 4]
2
-
out the components req
have to be discarded is sho uiring more than 35 cor
wn below by the shaded rections which = l- 0.8187 [ 1.2 2] =
1- 0.9988 = 0.0012.
area in the diagram.
Illustr atio n 34. A wh olesale distributor of fertilizer
for one type of fert products finds that the
iliz er is non nal ty distribu annual demand
of 16 tonnes. If he ord ted with a mean of 120 tonnes
ers only once a year, wh and standard deviation
is only a at quantity should be ord
5 per cen t cha nce of running sho rt? ere d to ens ure tha t there
Solution. (MBA, DU, /99 8, 200
Let the ann ual dem and (in tonnes) be don
0)
otcd by the ran dom var
iable X.
Therefore z = X- 12 0
25 16
Therefore, 35-25
z=
5 = 2.
From the tab le, the corresp ond ing value of z = 2 is 0.4
be .S - 0.4772 = 0.0228 772. The required sha
. Hen ce the num ber of proportion of the com ded arc~ will
thrown away is 2.28%. pon ent s wh ich will be
Illustration 32. A bak er has stud ied his record and notices tha
days in the yea r, the dem t for the pas t 310 worlcin&
and for his product (bread)
Demand ('000 units) has varied as fo\\ows
:
S 6 7 9 10
Number of days 8
What is the expected dem
20 60 80 \ 20 20 ,o
and for his product '? 120
lMBA. DU, J99.S\
 BUSINESS ST ATISTtc s
PROBA BILITY DISTRI BUTION S 521
520 ince the area between the mean and
· the figure. S
~
of O 05 is shown in bl we get this area of O•45 correspo nding Putting the value of np in ({)
The desired area •
-" · o 45 h rore from the ta e 10q = 36 or q = 3.6
the given value ot " is • • t ere,, , . d rrnal vanate.
• we get
to ;;; .. 1.64. . f - = 1 64 in standardize no Since the value of q is greater than one, there is some inconsis tency
in the statemen t given.
Subst1tuung this value o .. •
X-120 (b) Given 11p = 20 and ~npq = 4 or npq = 16 .
••• (ii)
1.64 = -,6
, )6
X - 120 + (1 ·64) (16) = .146.24.
-
or • •
then the wholesale d1stnbut or should order 147 20q = 16 or q = = 0.8 ; p = (J- q) = 0.2
·n whole units 20
If it is necessary to order I Putting the value of p and q in (ii)
tonnes. .. h t bulb will fail before 100 hours is 0.2. Bulbs fail
Illustration 35. The probabil~tyti t ~-tia lengths what is the probabil
ity that the number of
n(.8) (.2) = 16 or .16 n = 16 or n = 100.
independently. If 15 bub ls arc teste ord13~> '
failures betiore 1oo hours
•
. does not excceb assumed
• to follow binomia l distribut ions. - ~, PROBL EMS
Solution. The given problem can e
Herc n = 15 ; p = 0.2 ; q = 0.8. 1. Explain what is meant by Probability distribution of a discrete
random variable .
The required probability is given by : 2. Explain what do you understand by the term •mathematical expectat
ion·. How is it useful
P[X ~ 3) = P[X = 0] + P[X = l] + P[X = 2} + P[X = 3) for a business man ? Give an example to illustrate ils usefulness.
(MBA. Delhi Univ•• 1997)
= l~C (0.2)0(0.8)1S + ISC (0.2) 1(0.8) 14 + ISCi<0.2)2(0.8) 13 3
3. Define Binomia l distribut ion. Point out its chief characte
+ ISC 3(0.2) (0.8)12 ristics and uses.
0 1
12 • Under what
13
= (0.8) 15 + 15(0.2)1(0.8) 14 + 105(0.2) 2(0.8) + 455(0.2 )3(0.8) conditio ns it tends to Poisson distribution ?
(MBA. Osmani a. 1990)
= (0.8) 12 [(0.8)3 + 15(0.2)( 0.8)2 + 105(0.2 ) 2 (0.8) + 455(0.2 )3} 4. Define Normal distribution ? What arc the main characte ristics
of Normal Distnou rion?
= 0.0687[0.512 + 1.92 + 3.36 + 3.64} 5. What is Binomial distribution ? Under what conditions will it tend
to Normal distribu tion?
6. What are the chief propertie s of Normal distribut ion ? Describe
= 0.0687(9.432) = 0.648. briefly the importa nce
Illustration 36. A manufacturer who produces of Normal distribut ion in statistical analysis.
medicin e bottles, finds that O.l % of the (MBA. Delhi Univ. I 990)
bottles are defective. The bottles are packed in boxes containi ng 500 7. Under what conditions can observed (empirical) frequency distribut
bottles. A drug manufac- ions be ~pproxm 1atcd
turer buys I 00 boxes from the producer of bottles. to binomia l distribut ion ?
Usmg Poisson d1suibution. find how many boxes will contain : 8. What is Poisson distribution ? Point out its role in business decision
making. Under what
(i) no defectives conditio ns will it tend to normal distribut ion ?
(MBA. Delhi Univ. /996) (MBA. Kumaon Univ. 1998)
(ii) at least two defectives.
9. (a) Explain the term Random variable and Probability distribut
(Given e--0.s = 0.6065) ion of a random variable .
Solution. We are given : p = 0.001, n = 500, m = np = 500 x (b) Define the Binomia l variate and obtain its probabil ity distribu
0.001 = 0.5 tion function . Find the
mean and variance of the binomia l distribution.
(i) P[X =O}= e-m =e-0.S =0.6065
10. Discuss the distinctiv e features of the binomial , poisson and normal distribut ions. When
Therefore, the required number of boxes docs a binomia l distribution tend to become a normal distribution
?
= 0.6065 x l 00 = 60.65 or 61
(MBA, Shukhad ia Univ., 1995. MBA. Kumaou Univ. 2000)
(ii) P(X> 2) =1 -
[P,[X = 0) + P[X = 1)]
11. Under what conditio ns is the binomia l probabil ity model appropr
iate ? How does it •
= l - [e-<• + me....,} approac h the PoissQn probability model as a limiting case ?
= I - [0.6065 + 0.5(0.6065)} 12. (a) What is Poisson distribution ? Give example where it can
= I - 0.6065 {l .5} be applied.
(b) Explain binomia l distribution and give its applicati on in business
= I - 0.90975 = 0.09025. manage ment.
Therefore, the required number of boxes 13. Explain the meaning of Poisson distribution and state the conditio
ns under which this
= ]00 X 0.09025 = 9.025 OT 10. distribut ion is used.
Illustrat ion 37. (a) Bring out the fallacy, if any, in the followin
mc:111
g statemen t. "The 14. Explain briefly the characteristics of the Binomia l and Poisson
of a binomial distribution is 10" and its standard deviation is 6". distribu tions. How are
II. their means and variance s calculated ?
(b) The mean of a binomial distribution is 20 and the ·standard deviatio
p and q. n is 4. Calculatc
15. What is Poisson distribution ? Distingu ish clearly the relations
hip between the Bino-
mial and Poisson distribut ion.
Solution . (a) The mean of a binomial distribut ion in np and
standard deviatio n J;j;q · 16. (a) What is hypergeomctric distribotion ? Explain its properti
np = 10 and ~npq = 6 es.
Squaring npq = 36 (b) State the propertie s of the normal distribut ion.
 BUSINESS STATISTICS
522 distribution in business decision Fnakin
PROBABILITY DISTRIBUTIONS 523
. h . portance o f norma I 8,
(c) Describe briefly t e ,m 27 .The distribut ion of typing mistakes committed by a typist is
What are its chief properties ? . . given below· Assum-
f the normal probability distribution. Why dn- ing a Poisson model find out the expected frequenc
17. (n) Briefly describe the charac~n ies.
~uc~ o ~• Mistake s per page
· ent place in stausucs • O 1 2 3 4 S
it occupy such a promin . . . be a reasonab le approxim • •
ation of the binomial? No. of pages 142 156
(b) When can Poisson d1stnbut1on 69 27 5
(M. Com. DU, 1999)
(147.12 , 147.12, 73.56, 24.52, 6.13, 1.22] (MBA. Sulchadia Univ .. 1995)
·t ccidcnts lead to property damage of Rs. 100. Forty.._
F"fty per cent of all automo b t c a 28. In a normal distribut ion 7% of the observat ions
(c) •1 T cent lead to total loss, a damage o f Rs. 1,800. If a car h""r are under 35 and 89% are under
cent lead to damage of Rs.. 500. 63. What are the mean and the standard deviatio n of distribut ion
. en per cc·dent in a year. what is the expected value of the prooen.. IS ?
a 5 per cent chance of being 10 an a 1 (µ = 50.28, a = 10.25}
r-••1
damage due to that possible accident ? 29. If the average number of rejects in the manufac turing
. tt ry 000 tickets are sold at Rs. 10 each, and three pnzes •
arc to
process of a certain article
18. Suppose that m a 1o e 1 , •
is 4 per cent, what are the probabilities of 0, 1, 2, 3, 4 rejects in
a sample of 40 articles ?
. • television set worth Rs. 12,000, second pnze •
ts a short wave rad,·
be awarded. The first pnze ts· a (.6703, .2681, .0536, .00715,
worth Rs. , and the third prize is a cycle worth Rs. 1300. If you plan to buy • o .000715 )
1500 one ticket, wha{ 30. A Municip al Corpora tion had installed 5,000 bulbs in the streets of the city. If
is your expected gain or loss from the venture ? . . these bulbs have an average life of 800 burning hours, with a
_ Two investment opportunities arc open to prospective • standard deviatio n of 200
19 investor. If opportun ity A turns hours, find :
out to be successfu l, a profit of Rs. 6 lakhs will result and the probabil ity of A's success is estima (i) What number- of bulbs might be expected
as o 75, if A turns out to be a failure there will be a loss of Rs. 1 lakh.
to fail in the first 600 burning hours ? (ii) The
If opportun ity B sue number of bulbs expected to fail between 700 and 900 burning
a profit of Rs. 25 lakhs will materialize but if it fails hours, and (iii) the number
there will be a loss of Rs. 7 lakhs and the of bulbs expecte d to fail after 900 burning hours.
probability for B to tail is o 55. Which investment opportunity should
the investor take if the [i) 794, (ii) 1915, (i_ii) 1542.5}
dicision critenon is to maximize profits ? 31. The weekly wages of 1,000 workers arc normally distribut ed with a mean of Rs. t 700
(Opportunity BJ and a 'standard deviation of Rs. 150. Estimate that lowest weekly wages
20. If it rams a raincoat dealer can earn Rs. 5000 per day. If it is fair of the 100 highest paid
he can lose Rs. 1000 workers .
per day. What is his expectation if the probability of rain is 0.4 ?
32. Find the probabil ity that at most 5 defectiv e bolts will be found in a box of 200
(Rs. 1400) bolts if it is known that 2 per cent of such bolts are expected to
21. A firm plans to bid Rs. 3000 per tonne for a contract be defectiv e (You may take
to supply 1,000 tonnes of a metal, the distribu tion to be Poisson) .
It has two competitors A and B and it assumes that the probability that
A will bid less than Rs, (0. 784}
3000 per tonne is 0.3 and that B will bid less than Rs. 3000 per tonne is 0.7. If the lowest bid 33. If 20% of the bolts produced by a machine arc defectiv e. determin
gets all the business and the finns bid independently, what is the expected e the probabil-
value of the contract ity that out of 4 bolts (i) 0, (ii) 1 and (iii) at the most 2 bolts will
the firm ? be defectiv e.
(.4096, .4096, 0.9728]
22. If the probability that an individual suffers from reaction of a given
medicin e is 0.001 34. The mean inside diamete r of a sample
determine the probability that out of 2,000 individuals (i) exactly 3 of 500 washers produce d by a machine is
individu als (ii) more than 5.02 mm. and the standard deviatio n is 0.05 mm. The purpose
individuals will suffer from reaction. for which these washers are
intended allows a maximu m toleranc e in the diamete r of 4.96
l(i) 0.18 (ii) o.32) to 5.08 mm, otherwi se the
. washers are consider ed defectiv e. Determine the percenta ge of defectiv
23. If 2% of the electric bulbs manufactured by a company are defectiv e washers produce d
e, find the probabtl• by the machine assumin g the diamete rs are normally distribut ed.
ity that in a sample of I 00 bulbs (a) 0, (b) I, (c) 2, and (d) 3 bulbs will be defectiv e. (23.02}
[(n) 0.1353, (b) 0.2706, (c) 0.2706, (d) 0.1804] . 35. The probabi lity that any custome r who enters the store will
24. A certain type of plastic bag in the purchas e Colgate
past has burst under a pressure of IO pounds 30% of toothpa ste is 0.3. If 1.000 custome rs enter the store. what is the minimu m number of
the time. If a prospect ive buyer tests 5 bags chosen at random, Colgate toothpas tes the store must have on hand if the probabil
what is the probabil ity that ity that it will be out of
exactly one will burst ? stock is to be at most 1%·?
(0.3601 J
[334}
25. The probabil ity that A will make a profit on any busines s deal 36. Daily demand for a product
is 0.8, what is the is approxim ately normall y diustrib uted with mean
probability that he will make a profit exactly eight times in ten successi sales of 12 units per day and standard deviatio n of 4 units. How
ve deals ? many units must be on hand
[0.302) in the morning to assure no more than one chance in 5 of running
out of stock during the
26. One-fifth per cent of the blades p~oduccd by a blade manufac turing day?
factory tum out 10
be defective. The blades arc supplied in packets of 10. Use Poisson [ 16)
distribu tion to calculate lhe
approximate number of packets containing no defective, one defectiv 3 7. The probabil ity that India wins
e and two defective blades a cricket Test match against England is given to
respectively in a consignm ent of 1,00,000 packets. be 1/3. If India and England play three Test matches, use binomia
l distribu tion to find the
(98020, 1960.4, 19.6] prebabi lity that :
[MBA. DU. /999)
 BUSINESS STATISTICS
PROBA BILITY DISTRI BUTIO NS 525
524 ., ..) India will win at least one Test match ? 48. An automa tic deterge nt packing
(I') India will lose all three Test matches • (11 machine pr~.hlce s packag es whose weights arc nor-
" mally distributed with a mean of 8.00 gm and a standard deviatio
") 63 (ii) 0.7037) . b t Rs 500 on the outcom e of a roll or n of 0.010 gm.
I(81 An0 • 29 • (n) What is the probability that a package selected at
offered an op portun1ty to e of the
. • random will be between 8.00 gm and
individual ,s faces total seven or M, the
. . 8.015 gm.
a pair3 of dice. If the dice turn up so that the the sum
bet is lost. What is the expecte d value •
. . . (b) What proportion of packages are between 7.98
. Rs 1500 For any other outcome gm and 8.000 gm ?
individual wins • • (c) What proportion are between 8.005 and 8.151 gm
of the game for the individ~al 7 f 150 trucks is normall y distribu ted with.1a mean ?
J9. The fuel consumpuon 0 (d) What proportion are between 7,995 and 8.010 gm ?
of 8 n~et_ f km per litre. Use normal distribu tion to
of I 5 km per litre and a standard deviauo n ° 1•5. (e) What proportion arc above 8.017 gm ?
1 49. A manufac turer of electric
find the expected number o f trucks that 1·
average •
(b) 14 5 but less than 15.5 km per htre. • fuses packs fuses in boxes of 10 each and 2,000 such boxes
were sold. The previou s experie nce shows that 5 per cent
(n) 13 but less than 14 km per itre, • of the fuses are defecti ve. Using
Poisson distribu tion, find how many boxes will contain (i)
[(n) 24. (b) 39) . no defectiv e (ii) more than one
40. Fit a Poisson distribution to the data given below '' defectiv e.
x· o I 2 3 50. In a certain factory turning out razor blades. there is a small chance of 1/500 for any
fi 30 62 46 . 1 o
4
.2 blade to be defectiv e. The blades arc supplie d in packets
of I 0. Use Poisson distribu tion to
41. The heights of students in a class arc normally distribu ted calcula te the number of packets contain ing (i) no defectiv
wi~h a mean of 62 mches _an e Cii) one defecti ve and (iii) two
a standard deviation of 4 inches. What proport ion of the student defective blades, respectively in a consignment of I 0.000
s m the class have_ a height packets .
greater than 68 inches? What is the probability that a student 51. If on an average 8 ships out
selected at random wilt havc'a of 10 amve safely at port. find the mean and ~•andard
height between 58 inches and deviation of the number of ships arriving safely out of a total
66 inches ? • • of 1,600 ships.
42. One hundred car radio sets are inspected as they come [µ = 1280, CJ= 16}
off the produc tion lme and
number of defects per set is recorded below : 52. Assuming that sex ratio of male children is I , find the
4 probability that in a family of
No. of defects : 0 I 2 3
., 2
5 children, (i) all children wiJJ be of the same sex, and (ii)
No. of sets 2 0 three of them will be boys and two girls.
79 18 (i) 1/16 (ii) 5/16.
Fit a Poisson distribu tion to the above data.
[77.88, 19.47, 2.43, 0.2028 53. A compan y has 6 telephones which 10 executives use
, 0.1267 ) (MBA. Delhi Univ •• 1999) intermit tently. Assume that at
43. A machin e is suppos ed to drill holes any given time each executiv e has the same probabi lity 'p' of
with a ~iamet er of 1 inch. In fact the requirin g to use a telephone. If the
diameters arc normally distribu ted with a mean of 1.0 inches executi ves' requirem ents of telephones are indepen dent, the
and a probabi lity that exactly k execu-
standar d <ieviation of tives require a phone is b (k. n, p).
.02 inch. If there is a toleran ce of .02 inch, If on an average , an executi ve uses the telepho ne for 10
the holes should be betwee n .99 and 1.02 minutes per hour (p = 1/6), find the probability that 7 or more
inches. What percentage of the holes drilled are within executives need a telephone at the
toleran ce limits ? same time.
44. A hotel maintains two delux rooms. The demand for these rooms in any day is [0.0002 67}
distribu ted as a Poisson distribu tion with mean 1 .5. Calcula te the propor tion of day on 54. A machine produce bolts which are 10% defectiv
which neither of the rooms would be used ; and the proport e. Find the probability that in random
ion of days on which no demand sample of 400 bolts produced by this machine the number of
would be refused. defectives found
(i) will be at most 30;
[0.2231 , 0.1913 ) (ii) will be between 30 and 50,
45. ·The followi ng frequency table gives the distribu tion (iii) will exceed 55.
of 1,000 person s according
to their income : [(i) 22-8, cm 354.32, (iii) 3.56 l
Monthl y income (Rs.) No. of persons Monthl y income (Rs.) No. of persons 55. The mean and standard deviatio n for the life times of a population of light bulbs are
Below 5000 16 20000 -2500 0 166 1200 and 150 hours respectively. Assuming these lifetimes
arc normally distributed, what is the
5000- 10000 85 25000 -3000 0 100 probability that a light bulb-will last over 1500 hours?
10000 -15000 207 30000 -3500 [0.0228 )
0 69
15000 -20000 346 above 56. An [Link] of a publishi ng compan y, calculat
35000 l1 es that it requires 1 l months on an average
Fit a normal distribu tion to the above frequen cy table. Also to complete the publication process from manu~c ript to finished
determ ine the percentage books with a standard deviatio n
of persons with an income betwee n 17,500 ' and 27,500 of 2.4 months . He believe s that the distribution of publica
rupees. tion times is well describ ed by the
46. A car rental firm has two cars which it rents out normal distribu tion. Out of 190 books he will handle this
day by day. The number of year, how many wi11 comple te the
demand for a car on each day is distribu ted as Poisson distribu process in less than a year ?
tion with mean 1.5. Calculate
the proport ion of days on which neither car is used and [ 126)
the propor tion of days on which
some demand is refused . 57. An analyst predicts that 2.5% of all small compan ies will file for bankrup tcy in the
47. A dice is thrown 9,000 times and a throw of 3 or 4 coming year. For a random sample of 200 companies, estimate probability
is observe d 3,240 times. Show that
that the dice cannot be regarde d as an unbiase d one and (i) at least three wilt file for bankruptcy next year ;
find the limits betwee n which the
probabi lity of a throw of 3 or 4 is expecte d to lie. • (ii) exactly three will file for bankruptcy ;
 BUSIN ESS STATIST 1cs
526 PROB ABILI TY DISTR IBUTI ONS
will tile for bankruptcy.
527
(iii) nol more than five • (i) between 20 and 28 secs ;
lO 0 87 (ii) 0. l 45 (iii) 0.615 .
\] ber of accide ntal
5 •
~ P".st records show that the ~vera~e. ~um. h resort. If in drownmgs at a beach resort Is (Ii) less than 23 secs ;
I 00 000 of tounsls v1S1ltng t e a year 200,00 0 tourist s visited To what value should the mean service time be altered
3 nrr year for every • so lhat only I custom er in I 00 has
r T • thal • to wait longer than 50 secs ?
this resort, find l~e probabtdtt,es ·ng a~cident this year ; (I) 0.309 4 (ii) .4207, 26.7
(i) th re w,11 be no rowm
~ e will be al \east 6 acciden ts thi_s year ;_ 69. The marks obtain ed by the studen ts in an exami nation are known to be norma lly
(!!) there ·n be exactly 5 accidents th,s year distributed. If 10% of the studen
(111) there Wt
' ts got less than 40 marks while 15% got over 80. what arc the
than 8 accidents this year ~ mean and standard deviat ion of marks ?
• ) there will be more l ([Link] m.. ~/hi Univ.. 1989)
(IV 0 00279 (ii) 0.498 (iii) 0. \ 807 70. In a certain exami nation , 10%
(iv) 0.0~65 . of the studen ts got less than 30 marks and 97% of the
l(I) : b' •
59. Ftt a mom1a1 d,·stn'bution to the3 followmg 4
data • s~udents got less than 62 marks . Assum ing the
distrib ution to be norma l, find the mean and
X· 0 l 2 standa rd deviat ion of the marks.
/:. 28 62 28 12 46 [n = 42.96. a= 10.13]
{9.34. 40.48. ~5.78, 47 -51 1 6 • (~ 1 Airline s is having some proble ms with cash flows.
71. For a binomial distribution the mean is 4 and
variance 2. Find the probab ility of getting
60. The financial controller ~ a1 :J ult to predict whereas daily expen ses remain fairly
(1) at least 2 succes s
(ii) at most two succes ses
Daily revenue 0uctuate grca~~y an : ~ ;~eng ers If daily revenue has a normal distribution
c~nstant regardless of OOOthe da,/8~um :C~t of the valu~s [(i) 0.9648 (ii) 0.1445 )
with a mean of Rs. 72, lie below Rs. 82,000 , what is the stand
. . an per •
72. The follow ing table gives the numbe rs of days in a 50 day period in which autom obile
on of th e d'1stn'b ut·to n • What is the value above wh1c h 5 per cen t o f the val ues m
~
de\'1at1 • acc,de nts occurr ed in a certain part of a city. Fit
th: a Poisson distribution lo the data.
distribution lie ? No. of accide nts O l
. The following table shows the number of custom • h 2 3 4
61 ers retumm g t e produc ts m •
a mark, No. of days 19 18 8
mg territory. The data is for 100 stores : 4 l
(18.4, 18.4. 9.2. 3.1. 0.8)
No. of returns : 0 l 2 4 5 (M. Phil. K11rulcshe1ra. /991)
No. of stores 3 6 73. In a book the following frequency of mistak
4 l4 23 23 es per page was observ ed. Fit a Poisso n
18 9 9 distrib ution.
Fit a Poisson distribution.
No. of mistak es O 1 2
[4.97, 14.91, 22.36, 22.36. 16.77, 10.06, 5.03] 3 4 5
62. Three fair coms . . . . per page
are tossed 300 times. Find the frequencies of the d1stnbut1on of h
and tails and tabulate the result. Also calculate the No. of pages 630 160 90
mean and standard deviat ion of the distribu.. [463, 357, 137, 35. 7, 1) 70 30 20
tion.
63. How many workers have a salary above Rs. 2675 74. The distrib ution of monthly incom e of 4000
in the distrib ution whose with mean Rs. 6000 and standard deviation Rs. 1000 empol yees follow s norma l distrib ution
salary is Rs. 2400 and standard deviation is Rs. avera:
100 and the numbe r of worke find :
15,000, if the salary rs in the factory ij (i) Numb er of emplo yees having incom e more
of worker s follow s the norma l law ? than Rs. 7,000;
64. The Delhi Municipal Corporation mstalled 2.000 ' (ii) The numbe r of emplo yees having incom e
bulbs in the streets of Kai\as h Colony, less than Rs. 5500:
If these bulbs have an average life of 1,000 burning hours, (iii) The least month ly mcom e among
with a standard deviat ion of 200 hou~ the highes t paid 100 emplo yees.
what number of bulbs might be expect 75. (a) Proof readin g of 200 pages of a book contai
ed to fail in the first 700 burnin g hours? results : ning 500 pages gave the follow ing
l134}
65. In 24 trials of an event of small probability, the (MBA. HPU. /989) No. of mistakes per page : 0
frequency f of the numbe r of success X 1 2 3 4 5
1s given in the following table : Friequ ency 113 62 20 3 1 1
X 0 Cost per page
I 2 3 4 5
f 3 6 for check ing I .0
2 6 5 5 t 1.5 2.5 3.0 3.5 4.0
. . Th_e me~n number of succes ses is 2.75. Find 2 (a) Fit a Poisso n distrib ution.
d1stnbut1on with the same total frequency. the expec ted freque ncies of the Poisson
(b) Estima te the total cost of correcu,1g
the whole book. (MBA. M.D. Univ.. /995)
66•. The a_ppearing of 2 or 3 on a dice is connected as ((a) 109.76 , 65.85, 19.15, 3.95, 0.592, 0.0711 (b) 272.13 9)
and the tollowmg results arc observed : success. Five dice arc thrown 729 times
. 76. Which probabilily distribution is most likely
the appropriate one to use for the follow-
Number of ing data : Binom ial, Poisson or Normal ?
successes 0 \ 2 (i) The life span of a female born in 1957.
3 5
. Frequencey . : . 4~ 4 (ii) The numbe r of autos passing through
195 237 132 81 30 a toll booth.
Fit the
67 F' bmom1h
al d1strtbut1on assumi ng the dice to be b' d (iii) The numbe r of defective radios in a lot
of 100.
b • f iv~ undred. T.V. sets arc inspected as they un 1ase • (iv) The water level in a reservoir.
come off the produc tion line and average:
num er O s~ ects per set is found to be 0.402. Find 77. A book has 700 pages. The numbe r of pages
or more defects. the expected numbe r of T V sets having one with variou s numbe r of mispri nts is
• • record ed below . Fit a Poisson distribution to the
[165.5 ) given data :
Numb er of Misprints X 0 1 2 3 4 5 Total
68. T_he _time req_u1red by bank cashier to deal with Numb er of Pag~s with
a custom er has been observ ed to be
~;7'al~ y diSlnbute~ ~ th mean 25 secs : and a standa Xm1s pnnts 616 70
a a customer arriving at random will have to wait rd deviation IO secs. Find the 'probabilitY 10 2 1 l 700
: (M. Com, DU., 1999)