Probability Distribution

Thc set of values x; together rvith their probabilities P1 constitute the univariate probability distributiorr.

. The variate takes a discrete set of values then the distribLrtion is called a discrete probability
distribution or probabil ity function.
. The variate takes a continuous set of values then the distribution is called a probability density
f unction.

Bernoulli Distribution
A ranclom variable X is said to have a Bernoulli distribution with parameter p then thc probability
firnction is given by

P(x :,, : x= o,t (1)

[6':r ;rl)"';;,,{or
Theparameterp satisfies 0 < p < 1 and (1 -p) isdenotedasq.
\ randont erperiment whose outcomes are of two types, success .S and failure F, occuning with

probabilities p and q respectively,iscalledaBemoulli trail. lfforthisexperiment.arandomvariable

X is dellned such that it takes value I when S occurs and 0 if F occurs, then X fbllows a Bernoulli
distribution. \\Ihose mean E(x) : p and V(x) : pq . That is,
E(x)=0'q+1.P:P
E(x') = 02'Q -t 1''P - P
Var(x) : E(xz) - lr7x11z : p - p2 : p(l - p) = pq

Binomial Distribution
Let a random experiment be performed repeatedly, each repetition being called a trial and let the

occurrence of an event in atrial be called a success and its non-occurrence a failure. Consider a set of n
independent Bernoullian trials in rvhich the probability 'p' of success in any trial is constant for each
trial. then e : 'J,
- p , is the probability of failure in any trial.
The probability of x successes and consequently (n -x) failures in n independent trials, in a specified

order SSFSFFFS...FSF (say). Then by the compound probability theorem lve have
P(SSFSFFFS ... FSF) = P(s)P(s)P(r)P(s)P(r) ... ' P(F)P(s)P(F)
= p.p.q.p.q ....q.p.q = p.p.p ...p , q.q.q ...q = p*qn-*
L

BLrt x successes in ?? trials can occur in (|) ways and the probability for each of these \\'a)'s is same. viz..
p'q'-'. Hence the probabitity of x successes in n trials in any order is given by the aclditional theorenr
o1'probability by the expression (i)p,q"-r.
Definition : A random variable X is said to fbllow binomial distribution if it assumes only non-negative
values and its probability function is given by

p(x=x)=p(x)={(1)n'{-., x=0,L,2, "' n; q=7-p (2)

[0 , otherwise
The two independent constants n and p in the distribution are known as the parameters of the
distribution. Then any random variable which follows binomial distribution is known as binomial variate.

Shorv that total probability equal to one.

t7n

I n{r) = I (;) p'q'-' : ncop'q'-o * 'crp'qn-r * nczp2q,-z * nczp3qn-3 *

i=0

.... + nCnpnqn-n

: qn + npqn-r +f!j!f An-'+ ....+pn

: (p * Q), :1. (showed)

Properties of Bi nomial Distribution

l'he mean of the binomial distribution is given by
E(x) = H= L!=oxp(x) =O.p(x:0) + l.p(x:1) + ... .*np(x =n)
- 0.q" * l.nqn-rp *Z nCzqn-rpz * ...*npn
: nplQn-L+ (n - t)q"-rp * n-76rrtt-3o2 + ... t pn-t1
: np(q t p)-, : np (3)
Now, x2:x(x-1) + x
,. E(xZ) : Etx(x
- 1) + x)
= E{x(x - 1)} + E(x) (4)
Here. E[x(x - 1)] : ZT=ox(x - r)p(x)

= 0(0 - t)q" + 1(1 - l)nq"-, + z(z - 1) ,crqr-zp, + 3(3 - 1) nCzq"-3p3 *

.... *n(n - 1)p"

=n(n-l)qn-zrz *n(n- 1)(n -Z)qn-srz + ...+ n(n-l)pn

: n(n - 1)p' {q"-' + (, - z)qtt-2, + ... + p"-2} - n(n - I)pZ (q I p)n-z : n(n 1)p,
- (s)
From (3). (4) and (5), we get

E(x')=n(n-7)p2+np
'l'he variance. V ar(x) of the distribution is
Var(x) : oz : E(xz) - tE(x))' = n(n - l)p' * np - n'p' : np(7 - p) = npct (6)

Example:
Suppose that items coming off a production line are classified as defective or nondefective. l-he
probability of a defective piece is 0.04. I 0 items are draw'n at random during a short period anc'l classif-ied

according to the above scheme. What is the probability of (a) no defectives, (b) at least two defectives or
(c) one or two defectives?

Solution:
Here.n:!0andp:0.04.
So. q=1-p=0.96
(a) P(x - 0) : qLo : (0.96)10 :0.6648
r'ct P(.t > 2) : 1 - {P(x : 0) * P(x = 1)\

- 1. -[q', + tlq,p] - I - to.664r + 0.2770] -'1. -0.9418 : 0.0582

: P(.r = 1) * P(x :2):10qep * 4\qgpz = 0.2770 + 0.0519 : 0.3289

Erample:
A and B plal a game in which their chances of winning are in the ratio 3 : 2. Find ,4's chance of winning
at least three garnes out of the five games played.

Solution:
Let p be the probability that'A'wins the game. Then we are given:

n:5,-3 p: -2
+ Q= L-p:
u E

Hence, by the binomial probability larv, the probability that out of 5 games played, ,4 wins 'x' games is
given by

P(x : x) : p(x): (:) O'O'-' ; x:0,1.,2,...,s

The required probability that ',4' wins at least three games is given by:

p(x>3): Xi==(:)!;:
*[(:)r,+(i)" 3xz+1x32x1] = r"##!12=0.68
,l

Example:
A coffee connoisseur claims that he can distinguish betlveen a cup ol instant coffee and a cup of

pcrcolator coff'ee i5Yo ofthe tirne. It is agreed that his claim rvill be accepted if he correctly identifies at
5 o1-the 6 cups. Find his chances of having the claim (i) accepted, (ii) rejected.
r'vhen he does have
least

the ability he claims.

Solution:
If p denotes the probability of a correct distinction between a cup of instant coffee and a cup of
percolator coffee. then we are given:

o:#:1 = e:r-p:; and n:6

If the random variable X denotes the number of correct distinctions, then by the Binomial probability law,
the probability of x correct identifications out of 6 cups is given by:
P(x -- x) : p(x): (:) O' O'- ; x : 0,1',2, "',6
L: r The probability'of the claim being accepted is:

p(x >5) = p(5) + p(6) = (:) 0'0'-' * (:) (1)' = #+ #= 0 534.

,.ir The probability of the claim being rejected is:

P(X < 4) : L - P(X 2 5) = 1 - 0.534 = 0'466'

Erample:
A depaftmenr in a y,orks has l0 machines which may need adjustment from time to time during the day'
the day,
Three of these machines are old, each having a probability of 1/1,1, of needing adjustment durirrg
and 7 are new. having comesponding probabilities of 1'/21'
Assuming that no machine needs adjustment twice on the same day, determine the
probabilities that on a

particular day
(i) Just 2 old and no new machines need adjustment'

(ii) Ifjust 2 machines need adjustment. they are of the same type'

Solution:
Let pr : Probability that an old machine needs adjustment
1 = Ar: fr; and
=

that a new machtne need.s adjustment : * ) Qz = 2L

20
pz = Probabiltty

Then pt(x) = Probabiltty that 'x' old machines need adjustment

= (=,)pfqtr-* = (;) e)' L*J'-', * = o,!,2,3

4
and pz(x) = Probability that-x'new machines need. ad.justment

= (i)piql-' : (:,)(*)' (#)'-' , *: o,t, ,7

(i) The probability that just two old machines and no nerv machine need acljustn-rent is given (by the
compound probability theorem) by the expression:

P,(2) Pz(o) : (:) (;), (#)(1), = 0 016

(ii) The probability that just 2 new machines and no old machine need adjustment is :

pr(o) pz(z) : (#)'o (;)' (;)' : o o2B

The probability that'ifjust two machines need adjustment, they are of the same type' is the same as the
probability that'either just 2 old and no new or just 2 new and no old machines need adjustment,.
.'. Required probability : 0.016 + 0.028 = 0.044 .
/

Poisson Distribution
only non-negative varues a'd its
a poisson distribution if it assurnes
A random variable X is said to follow
probabilitY function is given bY

1e-^lx : x=0, \,2,....i 1>o

P(X- x)=P(x,7)=i
r\ [0,
Yt'
otherwtse
as the parameter of the distribution'
Here .t is known
employed:
Poisson distribution may be successfully
Following are Some instances where
(i)Numberofdeathsfromadiseasesuchasheartattackorcancerorduetosnakebite.
(ii) Number of suicides reported in a particular city'

tiii)Thenumberofmisprintsperpageinalargevolumeofcomparableprintedmaterial.
trucks'
rir ) Tl-re number of tyre failures per week for a fleet of delivery
of questionnaires'
1 \ r Trre number of responses per day to a rarge number
deaths per month in a large city'
,, r r The number of automobile
per minute during the busy hours
of a day'
,. 1r I T.e number of cars passing a crossing
a busy airport during an hour'
i r rrI he number of airliners arriving at

to one'
Shou that total probability equal

Lf="+ = e-) If=.f = s-).A - '

Properties of Poisson Distribution

Mean - E(x): II=o x P(x) - II=o.+

* **-\ * ....] = ... +ffi+ ]
= e-t [o.r + t.1 + ,.*. ... . .
^r-^{r+^+!+ (8)
-- 4e-tett - 1

Again. x? = x(x - t) + x then

E(x)z = Etx(x - 1)] + E(x)
-^^,
I}=o x(x - 1)P(x) : f,}=o x(x - D+
So, E{x(x - 1)i =
-- e-^[o + o +2(z-,)++ 3(3 -')*+ "'+ x(x -rlf + ""]
7

. "(
= e ^)"11 +)+;+A/ ...+,,,r+....
^'2
I _)^2 )
]: r-')'en =
,.2
(e)
^-
:. E(xz): - 1)] + E(x) : 12 + from (8) and (9)
Etx(x
^
Hence, Var(x): 02 : E(x') - {E(x)}2 : 1? + 1- 72 = 1
Thus the mean and the variance of the distribution are both equal to 2 .

Example:
A manuf'acturer of cotter pins knows that 5o/o of his product is def-ective. lf he sells cotter pins in boxes of
100 and guarantees that not more than l0 pins will be defective, what is the approxirnate probability that a

bor rrill fail to meet the guaranteed quality?

Solution:
Here rr = 100. and p : Probability of adef ecttuepin:5% = 0.05 (Since'p'is small, we rnay
,.se I)tri sson distribution).

.'. ). = Meannumber of def ectiue pins - np : 100 x 0.05 =5

et ihe randont variable X denote the number of defective pins in a box of 100. Then the probability of x
ie;ectir e pins in a box is

P(X : r): += .f i x: o, 1, 2, ....

Probabiiitl that a box rvill fail to meet the guaranteed quality is

P(x> 10)= 1 -P(x < 10) -',t - xigre-=- 1- r-tllgoI

Example:
An insurance company insures 4,000 people against loss of both eyes in a car accident. Based on previous
data, the rates were computed on the assumption that on the average 10 persons in 1,00,000 will have car
accident each year that result in this type of injury. What is the probability that more than 3 of the insured
u'il1 collect on their policy in a given year?

Solution:

Here n = 4000. and p = Probability o f loss of both eyes in a car accident : *#-
: 0.0001

Since 'p' is very small and n is large, we may use Poisson distribution. Thus the parameter .1 of the
Poisson distribution is

). = np: 4000 x 0.0001 = 0.4

Let the random variable X denote number ofcar accidents in the batch of4000 people. Then

7
r
e-A 1'
P(X-x)- "-o.+
19.4)x
Xl.

So the required probability that more than 3 of the insured will collect their policy is given by
P(X >3) : t- lP(X : 0) * P(X =1) +P(x :2) *P(x : 3)l

- - '). e-o n
{{o
+)o + (0.4) +9?:- !$}
- 1- 0.6703(1 + 0.4 + 0.08 + 0.0107) -',J, - 0.6703 x 1.4907 : 0.0008.

Example:
A manufacturer, lvho produces medicine bottles, finds that 0.1% of the bottles are defective. The bottles
are packed in boxes containing 500 bottles. A drug manufacturer buys 100 boxes fiour the producer of
bonles. Using Poisson distribution, find how many boxes will contain:
(i) no defective. and (ii) at least two det'ectives.

Solution:
He:e,\'=100,n:500,andp:Probabilityofadefectiuebottle=0.001,and7-np:500x
I001 = 0.5.
L:: ll-: r:rdonr rariable X denote the number of defective bottles in a box of 500. 'fhen the probability of
.r; .1:iectire bottles in a box is given by

P(X =r) = #:'!nTq ; x : o,t, 2,....

Hence in a consignment of 100 boxes, the frequency (number) of boxes containing x detbctive bottles is

f (x): N.P(x - x) = gETry

(i) Number of boxes containing no defective bottle

= 100 x P(X - 0) = 100 x 0.6065 = 61

(ii) Number of boxes containing at least two defective bottles

= > 2)l : 100[1 - P(X - 0) - P(x : 1)J

100 lP(X
: 100(1 * 0.6065 - 0.6065 x 0.5) : 100 x 0.09025 = 9.

8
7

Normal or Gaussian Distribution

'l-he
nrost imporlant probability distribution for describing a continuous randorn variable is norr.nal
probability distribLrtion. The normal probability distribution has been used in a rvide variety of practical
applications in r.vhich the random variables are heights and weights of people, test scores, scientif-ic
measurements. amounts of rainfall, and so on. lt is also widely used in statistical inference.
Definition:
A random variable X is said to have a normal distribution with parameters pr (called'rlean') and oz
(called 'variance') if its density function is given by the probability law:

f (x; tt,o) = ;*"*p{ ;(+)'l (ro)

- _J-r-{r-LD'/2o' ; _o(< x <x.,_c(< p <o( and o ) 0.

Importance:
1. \1ost olthe distributions occurring in practice, e.g., Binomial. Poisson. Hyper-geometric distribution
etc. can be approximated by normal distribution. Moreover, many of the sampling distributions. e.g.,
St',rdent's. Snedecor's F, Chi-square distributions, etc., tend to normality fbr large samples.

l. Eren if a rariable is not normally distributed, it can sometimes be brought to normal tbrm by simple
transtbrmation of variable. For example, if the distribution of X is skewed, the distribution of VI
rri'ht come out to be normal.

3. Manl of the distributions of sample statistics (e.g., the distributions of sample mean, sample variance.
etc.) tend to normality for large samples and as such they can best be studied with the help of the
normal curves.
4. The entire theory of small sample tests, viz., t,F,Xz tests etc. is based on the fundamental assumption

that the parent populations from which the samples have been drawn follow normal distribution.
5. Normal distribution finds large applications in Statistical Quality Control in industry for setting control

lirnits.

Properties:
I . The curve is bell-shaped and symmetrical about the line x= p.
L Mean. Median and Mode of the distribution coincide.
3. As x increases numerically, f (x) decreases rapidly, the maximum probability occurring at the point

x : lt, and is given by: lp(x)l*o, :

#.
4. fit:0 and Br:3 .

5. Ltzr+r = 0, (r : 0,1,,2,...) and pr, : L.3.5 ... (2r - 1)o2', (r : 0, 1,2, ...).
 the x-a'xis'
no pofiion of the curve lies belo$
6. Since l(x) beingthe probability. can never be negative'

T.I,irrearcotnbinationofindependentnormalvariatesisalsoanormalvariate.

to one'
Show that total probability equal
lf" '-(x-zVdx
tL)'

f *r,*ro*: -lorlfi J-" e

Substituting Y:x-lt = dY=dx

y2
2- ln o- ;7
: ffiJo' dy

Here 2ozz v'

ituttng,
in. substituti
Again.
y2
= ,A then fr= #.rt =+ #-- *
1

odz oz -Zdz
\
tnen --
dz o'[u 'lE -
= ---: dY=
- y= (2o22)2
dYo"o -
,[T, ,12

So' 1

r.r-i z.r-T dz
-= f(x)dx= #ffr-' ,r, oo = fi5t{"
- o dz = :h t:, -
I

+ f; e- ''z-r dz = +.0 = !n7'rfi = 1 (showed)

E(x) - #!i**u-I(*)' a* (11)

, = T ; E(z) = o and var(z) = 1'
(12)
OZ=x-p - x- 1t*oz

#:0*o + dx:odz
Then
_22

E(x) '-L1"fu*oz)sz dz; -o( < z 1X.

"-x-'
^lzt _22

dz *
-22 O 40(
zezdz
- +f**e,
,lzn "-x hJ_"
= p.1* o.E(z) = P+ o'0 = F
_zz
is an odd function of z')
(Since the integran O
'zn+tg-z-

10
 Moments of Normal Distribution
Odd order moments about mean are given by:

Itzn+L : ,f,c, - p)2n+L f (x) dx = #f*(x p)2n+Lexp{-(x

- - p)2 /zoz}dx
Using the substitution (l l) and (12), we have

-- h t:"toz)2n+lexp{-22 /z}dz = # t:*rzn+rr-22 /zdz = 0

Since the integrand ,rn*r94 is an odd function of z.

Er-en order moments about mean are given by:

t-tzn = [f, - p)2"f (x)d,x _ # t:-oz2n ,-'] a,

:..:rg t11) and (12))

-:n 22 ^2n z

-ufoo2Tl
^,2_
r_*, e-Td.z - ff .Z I{ ,r"e-'id.z
z2
b, T: f then

: # I{{zt)"r-'#_ '# t:, -t r("*))-, at

Hzn: T..("+) (1 3)

Substitute rt = L,Z in(13) we have

ttz:
Now putting
ff r(, * ;): #
n:2 then we have
i.0:#; .li - o2

ttq :'# .(, *) :'#.r{(r +i) *rj: # (r*i) r(r+i)

:#i;f H:#.ti:3oa
Hence Bt: fi: O and. Fz : 4 : E.

1.1
Note:
f(n+1) :n.t, fn: (n-1)f(n-1), r(n+1) :nf(n), f(0) :(, f(-i) - -o(,
f(-n) = -c( , aO : .li, r(1): t, r(n) : f{x"-le-'dx , {"e-*'dx: ),fi
pr: Jx f (x)dx , tti: jxz f (x)dx , pL: Ix3 f@)ax , Fz= pz- Qt)2 ,

ltz : ttlz - 3ltzttt + 2(p)2, [t+ : lt+ - 4ttzltr * 6pzjt)z - 3(ltr)n

Median of Normal Distribution

If M is the median of the normal distribution, we have

[!*f(.,oa*:) + + l''' ,- ruL,

L rM -@-p)2
otlztt u-x 2
L

L rLt -tx-412 1 , (x- tt)z

' l'e zo2 dx* --= "M e--iT
l" O*
1-
(14)
olzrt J-x olZn "U 2

- !
22
B,r #l:-r-\#d*:GI: e-vdz 2

Then *e get fiom (14)

t+ -l r-E#d"x _
olZrt t*
1
2 11 " 2

+ -+ f' ,-\#dx -
o{Zn'U
o

:. !l- M
Hence for the normal distribution Mean : Median.

I\Iode of Normal Distribution

\lode is the value of x for which /(x) is maximum, i.e., mode is the solution of
f'(x) = o and f" (x) < o
For normal distribution with mean p and standard deviation a2 ,

togf(x)-c-fi{*-D,
where C : log(7/ orln) is a constant. Differentiating w.r.t. x , we get

12
 I

*, f'(x)- -#(x-rD
+ f'(x)--#(x-tDf@) = (x - F) : o

Again, - -;(x - tD f @)
f '(x)
, f" (x) - -*V.f(x) + (x - tD f'(x)l

Hence x: p is the mode of the normal distribution'

That is, the
Theretbre. rre find that for the normal distribution mean, median and mode coincide.
: rstr-ibution is s1'mmetrical.

Erample:
X is normalll distributed and the mean of X is 12 and S.D. is 4. Find out the probability of
the follon'ing:
(a) (i)X>20, (ii) x <20, and (iii) 0<x<12 (b) Find x,,when P(x> x,):0'24
(c) Find xi and xi . rvhen P(xo < f . *r): 0'50 and P(X ' *r) = 0'25
Solution:
(a) We have trr - 72, o = 4, i.e., X- N(L2, L6).

(i) P(x 2 Z0) =7

When X=Zo,Z-20-L2-
4
2

:. P(X > 20) = P(Z 2 2) = 0.5 - P(0 < Z <2) = 0.5 - 0'4772 = 00228

Z=O
X=lt 13
(ii) P(X < ZO) : L - P(X > 20) - 1. - 0.0228 = 0'9722
(iii) P(0 < X < 1'2) = P(-3 < Z
=0),
(z: "#)
= P(0 <Z < 3) = 0'49865

x'-L2 :
(b) WhenX= x' 27 , (say).
Then, we are given:
P(X>x')=0.24 + P(Z > zt) = 0.24 + P(0 < Z 1z) :0.26
From Normal Tables, 4 = o.7L (approx.)
0.26

X:F X:X'
Z=0 Z: Zr
*'-"=0.7L
Hence 4 = x'=12*4x0J1 =1.4.84'
(c) we are given : P(xo < x l xr) = 0.50 and P(x > x') = 0.25 ( l5)

From ( 5), obviously the points x[ and. x| are located as shown in following adjoining
1

When X = xi, Z : \!: r, (say) andwhen X - x'0. Z = +- -zr

X=x'o X= lt X=X|
Z=_Zt Z=O Z=Zt

We have F(Z>zr)=0.25 + P(0 < Z < zt) = O-25; "' Zt = 0'67 (Obtain fromTable)
Hence
xL-72
4
: 0.67 + xl=12+4x0.67 =14.68
and
*L -.72
4 - -0.67 - xl= t2-4x0.67 :g.32 -

1.4
Example:
X is a normal variate rvith mean 30 and S.D. is 5. Find the probabilities that
(i) 26<X< 40. (ii) X>45, and (iii) lx-301 >5.
Solution:
Here H :30 and o :5.
26-30
When X:26, Z:x-& - = -0.8 and when X:40. g -
a0-,3o
oss - 2

"'P (26 < x < 40) :'rt;Z'r! oiu==r1r==

r;,'l ;,oot=- 3.) Jri E ; -i;r2 o.r 6s3

X=26 X=F X=40

Z=-0.8 7=O Z=2

\\hen x=45, Z=ou-3':3 5

,', P(X > 4s) : P(Z 2 3) : s.5 - P(0 < Z < 3) : 0.5 - 0.49865 = 0.00135

X=45
Z=3

P(lX -301 < 5) : P(25 < X <35) = P(-1 < Z <L)

= 2 P(0 < Z < L) = 2x 0.3413 = 0.6826
:. P(lX - 301 > 5) = t- P(lx -301 < 5) = L-0.6826=0.3t74.

15
Note: lX-301 <5 + +(X-30)<5;
Taki's'vesign,lvehave X-30 < 5 = X < 35
'faking-vesign
-X+30<5 + -X<-25 3 X>25

Example:
kilos. Assuming normal distribution, how
The mean yield fbr one-acre plot is 662 kilos with a s.d. 32
yield
many one-acre plots in a batch of 1000 plots r.vould you expect to have
plots'
(i) over 700 kilos, (ii) below 650 kilos, and (iii) what is the lowest yield of the best 100
Solution:
lf the random variable X denotes the yield (in kilos) for one-acre plot, the we are given that
X- ,^/ (p, o').rvhere tt = 662 and o = 32 '
rir The probability that a plot has a yield over 700 kilos is given by

P(X >700) = P(Z > 1"1g),' :T\7

: 0.5 - P(0 < Z <1'19) : 0'5 - 0'3830 = 0'1-17

Hence in a batch of 1000 plots" the expected number of plots with yields over 700 kilos is
1000,0.117-111.
(ii) Required number of plots with yield below 650 kilos is given by:
650- 662
1000 x P(X <650) = 1000 x P(Z <-0'38), L_
5a

<
= 1000 x P(Z> 0.38) = 1000 x [0'50 - P(0 Z <0'38)]
= L000 x (0.5 - 0.1480) = 1000 x 0'352 = 352 '
(iii) The lowest yield, say, xtof best 100 plots is given by : P(X ) xr) = ffi = 0'1 '

When X = xr, Z =T : ry = zt (say), suchthat

P(z>zt)=0.! + P(0<2321)=o'4 + 4:l'28 (approx')

Thenxt:662*3221=662*32x1'28=662++0'96=70?'96'
Hence the best 100 plots have yield over 702'96 kilos'

16
 t

Erample:
electric lamps in the streets of the city' If
these lanrps
The Iocal ar,rthorities in a ceftain city install 10000
200 hours, assunring normality,
har,e an average lif-e of 1000 burning hours with a standard deviation of
fail (i) (ii) betr'veerr 800 and
in the t-rrst g00 burning hours?
what number of lamps might be expected to
of burning hours would you expect that (a) l0% of the lamps
1200 burning hours'i After what period

would f'ail ? (b) 10% of the lamps would be stillburning?

Solution:
bulb in burning hours' then we are given that
X- N (t'' o')'
If the variable X denotes the life of a

rvhere F = 1000 and o -- 200'

800 burning hours is given by
(i) The probability'p'that bulb fails in the first

p: P(x< 800) : P(z < -1) -- P(z > 1); ('= *#*)
:0.5 -P(0 <Z < 1) = 0'5 - 0'3413 = 0'1587

outofl0.000bulbs,numberofbulbswhichfailinthefrrst800hours
= 10,000 x 0.1587 = 1-,587 '
< z <1) : zP(0 < z < 1)
tiii The required probability = P(800 < x <1'200) = P(-1
=2x0.3413:0'6826

z=_L z= t
:=t lif-e is
Hence the expected number of bulbs with
life between 800 and 1,200 hours of burning

10,000x0.6826=6,826'

we have to tlnd x1 such that

(a) Let 10% of the bulbs fail after x1 hours of burning life' Then
( l6)
P(x<xr)=0.10. When x=xt -zt(say)
'=#=
zr) = O.l0 + P(0 < Z 121): 0'40
(17)
:. p(Z a -rr)= 0.10 + P(Z >
Thatis, P(-]'zB<z <1'28) = 0'80 + 2P(o <Z
< 1'28) = 0'80

17
 = P(0<Z<1.28):9.49 (18)

Fronr(17)and(18). rve get 21:1.28; .'. -'J..28 ) xt= 1000 -256:744

"#!9:
Thr.rs. after 744 hours of burning life. I 0% of the blubs will fail.

x,
X= X=F Z=zt
Z=-2" Z=O

tb) Let 10% of the bulbs be still burning after (say) x2 hours of burning lif.e. Then rve have

P(X> xz):0.10 =P(Z)zz)-}.rc; (22: t#ry) + p(0< Z<rr):0.40

:. zz = 1.28 ; i.e, !z:!!! : l.2B :e xz = 1000 + 256 = 1,256

Hence. alrer 1.256 hours of burning life, l0% of the blubs will be still burning.

X=lt X=Xz
7=O Z= Zz

Example:
If the skulls are classified as A, B and C according as the length-breath index is under71,
between 75 and 80, or ouer 80, f ind approximately (assuming that the distribution is normal)
the mean and standard deviation of a series in which A are 5Bo/o, B are 3\o/o and. C are 4o/o
^

being given that if

f (t) : h li exp(-xz /2) d.x , then /(0.20) : 0.08 and f (1.7s) : 0.46 .

18
!,,iuIicrn:
L-i lhe length-breadth index be denoted by the variable X, then

P(X < 75) = g.5g and P(X > B0) : 0.04 (le)
Since P(X < 75) represents the total area to the left of the ordinate at the point X = 75 and. p(X > B0)
Represents the total area to the right of the ordinate at the point X:80, it is obvious from (19) that
points X : 75 and X = B0 are located at the positions shown in the following figure.

0.08

X=lt X=75 X=80

Z=O Z=zr Z=z.t

J^ exp(-x'/2)dr represents the area under standard normal curve between the ordinates at
--ll'u
.,

Z = 0 and Z: t, Z being N(0,1) variate.

, f(t) = h$exp(-xz/z)dx:p(0 <z<t)
Hence [email protected]) = P(0 < Z <0.20) : 0.08; and)
(20)
f (t.75) = P(0 <Z <L.75):9.46 )
Let pt and o bethemeanand s.d. ofthedistributionrespectivery. Then x-N(tt,o2).
When X:75, ,:1:-: z, (say), and when X=80, Z=9!=: z2 gay).
Thenfromthefigure, itisobviousthat P(X < 75):0.58 :r p(0 < Z <rr) = 0.08
.'. Using (20). we have z, :':;n : 0.20 (21)

Also P(X >80) : g.g4 + P(0 < Z t rr) : 0.46

.: Using (20), ive have z2:*=: l.7S (22)
Solvin_e the equations (21) and (22), we have

lt = 74.35 and o = 3.23 .

19