CHAPTER

2 Random Variables and

Mathematical Expectation

2.1 Random Variables

In the preceding chapter, we have seen that in many applied problems involving
probabilities like the tossing of a number of coins etC., we are rather more interested in
the number of heads or tails that appear, instead of the outcomes like getting heads or
tails. Likewise, when we roll a pair of dice we are generally interested only in the total,
and not in the outcome fo each die.

If we consider an experiment of tossing two coins, we get the following sample points or
Outcomes.

HH HT TH TT

Now, if we are interested in the number of heads that appear, we can always associate a
numerical value to each sample point 2, 1, 1 and 0. These numerical values are random
quantities that are determined by the outcomeof the experiment. Thus, each outcome of
arandom experiment can be expressed by a realnumber. The numerical value associated
with the outcome of a random experiment is called a random variable. In the above
example, if X represents the number of heads then, we can see that it is a random
variables defined on the sample space S given as,
S= {HH, HT, TH, TT}
n simple words,a random variable is a functionthat associates a real number with each element
tne sample space. Also, we can see that, here, X assumes values 2, 1, l and 0 with some
specific probability.
For e-g P(X =2) = P(HH) =l/4
 Probability
76
measure andX is a real-valued function defincd over th
space vitha probability
fSis asanple random
variables.

clements of S, then
X is called a
function also. IfE is a random experiment and S
random variable
as a
We can define a X(s), where s e Sis called a random variabls
associated it, a function
with
sample space or stochastic variate also
variate
called a chance
is sometimes
Arandom variable letters and their
random variables are denoted by capital
to denote a value
valuesh,
It is to be notedthat For example, we will use x
case letter.
the corresponding
lower that it is the set of elements
mentioned earlier P(X =2) means
random variable X. As X takes on the value 2.
random variable
which the
the sample space for
consider, may either be adisCTete sample space or a contiy
The sample space that we number of possibilities or an
contains a finite unendi
sample space. If a sample space called a dio
elements as there are whole numbers, then it is
sequence with as many numbera
On the other hand, if a sample space contains an infinite
sample space.
possibilities equal to the number of points on a
line segment, then, it is called a continuo
sample space.
variables can be classified as discre.
Depending on the type of sample space, the random
variable is called a discrete randor
or continuous random variables. Thus, a random
ofX. h
variahle if, it assumes only a finite number or countably infinite number of values
other words,it is a real valued function defined on a discrete sample space. A randon
variable is termed as a continuous random variable if it assumes any value in some interval a
intervals.

2.2 Discrete Probability Distributions

It immediately follows from the definition of discrete sample space, that the probabilt
measure defined over such a sample space automatically provides the probabilities thau
random variable will take on any given value within its range.
Forexample, in rolling two dice, if Xis a random variable
denoting the total (i.e., sumo
both the dice) rolled, then the probabilities associated
with all DOssible yalues of X
shown in the following table :
X =* 2 3 4 5 6 7 8 12
10 9
P(X =)| 1/36| 2/36 3/36
4/36| 5/36 |6/36 5/36 2/36| |/36
4/363/36
2.2.1 Probablity Mass Function
Most of the times, it is
variable Xby aformula.convenient and easy to represent all the probabilities of a randor
numerical valuesx that canNecessarily, such a representation
be denoted by f (), would be a function oft
gr) etc. Therefore, we
f(x) = P(X = *) can write
 77

in the above table, P(X =5) =4 /36

For .g..
Thus, the
difference values of the discrete random variable X
probabilities give us the probability distribution of together with their
associated
the random variable X.
Now the probability. mass function or probability function of a diserete
as the set of ordered pairs [x. random variable X
can be defined f()| if for cach possible outcome x, f()
satisfies the following conditions:
|1. f(u)20

|2. f)=l

The probability mass function is abbreviated as p.m.f.

(ILLUSTRATIVE EXAMPLEs)
Illustration l:A shipment of 8 similar fridges to a retailer contains 3 that are
a hotel makes a random purchase of 2 of these fridges, find the defective. If
probability distribution for
the number of defectives.
Solution :IfX is a random variable whose values xare the possible
number of defectives
purchased by the hotel. Then, x can have any of the values 0, I and 2.
Obviously,

10
f(0) =P(X =0) = 28

15
f() = P(X =I) = 28

3
f (2) = P(X =2) = 28

Thus, the probability distribution of Xis given as follows:

10 15
P(X = r) 28 28 28
 78
Probabl Rar

Whstration 2: if d coin ix tosxsedfo times, fnd the probability distrihution for gett
The

heads.
gettinga head or a tail is equal and
Selvtien As we know the probability of 2.

PH-P-; 3
the number of heads then
Thus, if acoin tossed twice. and Xdenotes
f(0) =P(X =0) =
1
f) =P(X=)=;
f(2) = P(X=2) =4
Thus,the probabilitydistribution of X is given as follows :
X=

P(X = r) 2

2.2.2 Discrete Distribution Function

Manyatimes we are interested to know the probability that the value of random variaok
X is less than or equal to some real number x. If F(r) = P(X 2), then F(r) is defined s
the cumulatie distribution or distribution function of the random variable X.
Thus, if X is a discrete random variable, the function F(x) given by
F(x) = P(X sr) = )f(0) for -o <*<o

is called cmulative distribution of X, wheref(t) is the value of

the probability of Xat t.
Lus assune that the random variable X takes values X),
),...., x, with probabil1te
P- P2 P and let A < ) <Iy <.... x then it is easily seen
that
PX < 4}=U, P[X ]=A. P[X< y]=P
PX S }= P{X <]+ PX =
Likrie, *)l=AtP

Thus, if we are given the

probability
corresponding function is generally easydistribution
to find.
of a discrete randonn
variable, t
 Random Varlables and Mathematical Expectation
79

The following properties of F(r) follows from the axioms of probability:

0s Fr) sI

2 Fu) is amonotonie inereasing function of xie.,

ifb2a then F(b) > F(a)
3 F(- e) = lim Fr) =0
F(o) = lim F(r) =0

4. FIb) -Fa) =P[Xsbj-P[X sal

= Pla<*sbl
5. F(a<r<) =P(a<xsb)-P(r= b)
= F(b)- F(a)-P(r =b)
P(asxsb) = P(a<xsb)+P(r= a)
= F(b)- F(a) + P( =a)
P(asr<b) =P(a<x<b)+ Pr= a)
= F(b) F(a) + P(r = a)-P(r=b)
6 F(r) has n points of discontinuity at x1, x), ....,. Xy. At
each of these points the
function is
continuous on the right, but discontinuous on the left.
7. At the point x, F(x;+0) = F(x; -0) + p, where F(I;-0) denotes
ari
the limit of F(x) as x
ned approaches x; through values less than x; and F(x; +0) denotes the limit when x
approaches x, through values greater than x

THEOREMS
Theorem 1: The range of a random variable X consists of the values
bil
4< < 3 <.... n, then

fug)= P(X =*)= Fr)-F(j-1)

Proof: We know,
F(x) =P(Xsx)
-2P(X=)-Z/u)
j=
and
Fxj-)= P(X Sx)=Z/)

le.
F(«) - F(x) fu)
 80 Probabilh
Thus, given the distribution function of discrete random variable, We can
compute
probability mass function.
Illustration 3: Find the distribution function for the following probabiltity distributio,
3
X* 2
6 4
4
P(X = *) 16 16 16
16 16
6 4
4
Solution :We are given f0) -f0
16 =s2)=G)16
It immediately follows that
F(O) =f(0) =
5
F() =f0) +f) =16
F(2) =f0) +f) +f2) =16
15
F(3) =f0) +f) +f2) +f3) = 16

F(4) =f(0) +f) +f(2)+ f(3) +f(4) =1

Hence,the distribution function is given by
for X<0

1/16 for 0sx<l

5/16 for Isx<2
F(r) =
11/16 for 2<x<3
|15 /16 for 3sx<4
for x4

lluyfation 4:Arandom variable Xhas the following

probability function.
-2
P(X = x) 2 3
0.1 k 0,2 3k 2k 0.3
(i) Find k
(ii) Obtain P(X < 2),
P(X s2), P(-2 < X <2)
(iii) Find the minimum
value of ksuch that P(X
Solution: (i) The definition of l)> 0.32
3 distribution gives us
t=-2
oba
Random Variablcs and Mathematical Expectatlon
81

0.1+*+0.2 +3A +2k+0.3-|

1hus,

but
PN2) =P(X =-2)+ P(X=-l)+P(X =0) +P(X - )
-0.1+*+0.2+3k+4k +0.3
17
30

P(X 22)=|-P(X <2)

17 13
30 30

P(-2 <X<2)=P(X = -l)+P(X =0)+ P(X =I)

=k+0.2+3k =4k +0.2+=+
30
P(X sl) =P(X =-2) +P(X =-l) +P(X =0)+ P(X =1)
=0.]+k+0.2+3k =4k+0.3
(ii) Asdesired,
P(X Sl) >032
4k +03+>032
4k>0.02
k=0.005

IMustration 5: Froma lot of 10 items containing 3 defectives,a sample of 4items is drawn

at random without replacement
(i) Find the probability distribution of X, the number of defectives
() Find P(X sI), P(X 2 I) and P(0 < X <2)

(i) Asample of 4items fromn a lot of 10 items can be drawn inways which are
3 equally likely. The number of ways in which out of 3,r defectives can be chosen and
0.3 out of the remaining7 items, 4-x non-defectives can be chosen
immediately follows that
P(X =A
Thus,
2
-

P(X = )
30
 82

s) =P(X =0) +P(X =l)

(ii) PX ,1_2

PX 2)=2 10 30 6

PO <X<2) =; distribution
function can be presented
cumulative
function andthe by two methods
probability mass can be represented
The function
graphically also. The
probability mass
and the other a bar chart. If we
have the following
probability histogram
being the
probability distributions
3 4
2
X =x 4
4 6
16 16
PX =) 16 16
16
shown in Fig. 2.l and Fip
chart can be drawn as
then, the probability histogram and bar
2.2 respectively.
f(X)
6/16

4/16 4

I/16

X
2 3

ig 2.1
f(X)
6/16 4

4/16

I/16

(
3 4
Fig. 2.2
 Variables and Mathematical Expectation
Random

83

a
prbabilhty
histoçram, the hcight of cach rretangle equals the prohability that
In casc of variable takes on the value that correspnds to the midpoint of its base. In
random

the
of har chart
also we have the same thing except that there is no needto have a
pontinuous horizontal scale

a umulative distribution appears to be astep function. Asimple exarmple is

Thc graph of
shownin Fig.
2.3. The graphis obtained by plotting the points (r. Fr).
FX)

3/44

i/24

|/44

X
2 3 4

Fig

2.3 Continuous Probability Distributions

random variables, we have seen that the
In the earlier section,while discussing discrete
tabular form. But in case of continuous
probability distributions can be presented in a
random variable has a probability of
randomvariables, this is not possible. Acontinuous
assuming exactly any of its values. As a result, its probability distribution cannot
zero of
assigning probabilities in connection with
De given in a tabular form. The concept of understood by
continuous random variables can be
Coitinuous sample spaces and weights of
example. Let us discuss a random variable whose values are the
Cissdering an kilograms, or say
any two values, say 2.9 and 3.0
bbies belw 3 years of age. Between which is 3
3.01 kilograms, there are an infinite number of weights, one of
2.) ad
probability of selecting a person at random who is exactBy3 kilograms and
ligans. The
large number of weights so close to3 kilograms that one cannot
A eo the infinitely zero to the
measure is of remotr chance, as a result we assign aprobability of
Umanly kilograms
alou the hances of seiecting a baby who is at least 2
F . Hwever, if talk than a point
ilograms, then we are talking of n interval tather
nd ied ore that 3
value of ur random varialbie. ranlom variable.
that the weights of lbabies is a cominuS
S interesting to note semest tenth ol akilogram it is a
case ofdiwrete
asmounts
wever, íf he rounds the distibution shown in Fig 2.4 If
again it is
a probability
Tandom variable which bas kilogam, again we deal with a
discrete random
hundredth of a
Tounded to the nearest
distribstion as givenin Tig 25.
Itwe keep ounding the
probability
variable which has a
 84 Probas
ten-thousandth of a kilogram, the
weights to ncarest thousandth or
probability distributions of the cornesponding discrete random variables histogram
and the sum of the areas of the
the ontinuous curve shown in Fig. 2.6 rectangles h,
amount falls within any specified
represent the probabilitythat the
curve.
interval approa,
the corresponding area under the

2.9 3.0 3.1

Fig. 2.4

2.90 3.00 3.10

Fig. 2.5

3.9 3.0 3.1

Fig. 2.6

2.3.1 Probability Denslty Functions

The definítion of probabílity in the continuous case presumes for each random Variabk
the existence of a function, called a probability density function, so that areas under
curve give the probabilities associated with the corresponding intervals along
horizontal axis. A probability density function or simply density function, integrate
from a tobwhen asb, gives the probability that the corresponding random variable w"
 Random
Variables and Mathemnatical Expectation
85

inthe interval atob. The density function or p.d.f. is generally denoted by

take on a value
the random variable X,
fr) tor
Thus, a function with valuesf(r), defined overthe set of all real numbers, is called a
prabability density function of the continuous random variable Xif and only if

PasXsb) =|f) àr
with a<h
for anyreal constants a and b
lisof importance to note that f (e), the value of the probability density of Xat cdoes not
dive P(X =c) as in the discrete case. In continuous random variables, probabilities are alvays
Asociated with intervals and P(X =c) = 0. As a result, the value of a probability density
function can be changed for some of the values of a random variable without changing
the probabilities.
Hence, if Xis acontinuous random variable and a and bare real constants with a b, then
PasX<b)= P(asX <b) =Pla< Xsb)
= F(b)- F(a)
Afunction f () can be called as a probability density of a continuous random variable
Xif itsvalues satisfy the following conditions :
1. fr) 20for

2.

Some typical density functions are given below :

f(X) f(X)

Fig. 2.7 Fig 2.8

f(X) ‘ f(X)‘

Flg. 2.9 Fig, 2.10

 86
Probab
It immediately follows fronn the propertics of a density function that a proba
density tunction is constructed so that the area under its curve bounded by the xa
range of Xfor which f()
is defined.
computed over the
whenXtakes a value betwecn a and b is equal tothe shaded arca undt
cqual to l that
pobability

ordinates at x =a and x =b and

is shown Fig, 2.11.
density function between the

f(X)

Fig 2.11

The shaded area in Fig. 2.1l gives us P(a< X<b)

Mustration 6: Find whether the following function is a probability density function:
-l<x<2

|0 elsewhere

Also obtain P(0 < X <l) [C.CS.U.M

2
Solution: Here,

Hence, f(x) is a p.d.f.

Now,

Illustration 7 :IfX has the probability

density
flr) = k.e-3* for x>0
elsewhere
find k.

Solution:For f(x) to be a p.d.f.,

we must have
 Variables
and Mathematical Expectation
Random
87

=k lim - k=3
H e nw eh a v e ,

&:1fa random variable X

has the probability denslty function as follows :
Mastration

f()=-2<r<2
=0 otherwise

values of
Obtatn the
(ii) P(|X|> 1) and (ii) P[(2X +3) > 5].
(6) P(X <l)

Solution : (i)
-2

() P(|XM>) =l-P(| X\<)=1-Jf(dr

-1-2 4 2

+3) >5] =P(2X >5-3) = P(X >) =l-P(X<l)

(ii) P[(2X
3 1
4 4
function:
that the following is a probability density
Ilustration 9: Verify
-3S r s-1

-6-2x. -Isrsl

"0-t. 1srs3
where the variable Xis in the range (-3, 3)
Solution : Here,
1 3
3

-3 -3
 Probabih
3

defined asfollows a density function :

lustration 10: Is thefunction
fr) =0 x<2

2srs4
-+21)
=0
having this density will fall in the
interi
probability that a variate
Also obtain the [C.CS.U.
2sXs3.

Solution : Here,

fr) 20 r in the given interval and

2 4

2 4

This implies that f (r) is a density function.

Now,
3
4
P(2 SX<3) =J+2)dt =
2

llustration ll : Verify that

f)=x-srso is a probability
I+
dent

function.
Solution :We see that for all points xin -oo Sx < oo, f(x) 20

and

Hence f () is a probability density

function.
 Random Variables and Mathematical Expectatlon
89

llstration 12 : Find whether its following is a probability density function

-l<r>2
elsewhere

Also obtain PO
< si) {C.C.S.U. 2012, 05]
Solwtion : Given

f(*)=-l<r<2
=0, elsewhere

It can be easily seen that f(r) 20

Now,
2

2
2

=0+ + 0

Fience, it is verified that as f )20& fr) dr=1, fu) is aprobability density

function.
Also,

0
llustration 13: Is the following function f (*) a density function?
s/2
f(u) =I/2 l<rs2

|3/2 -*/2 2<*s3

Solution: As we know.

f(r) willbe ap.d.f. if (1)f() 20 and (ii) |f() dr :

Here we see,

-20 for 0srsl

;>0 for I<rs2

3_0 for 2Sx <3
2 2

Hence the condition f(r) 20 is satisfied.

Now,
0 2

4 2 4

Hence the conditionf) dr =l is also satisfied.

As both the conditions are satisficd, we can say that f(x) is ap.d.f.

2.3.2 Varlous Measures of Central Tendency, Dlsperslon, Skewnes

and Kurtosis for Continuous Probablity Distributlons
defi

Let f(r) be the probability density function of arandom variableX whereXis

from a to b. Then,
 ndom Varlable and Mathematical xpectation

Arahmetic mean: Thit it einmly given by

AM-f:fima
Harmoni mean: 1The Harmnik mean His given b

Coometric mean : Geometric mean (G is given by

4 R, (rth moment about origin)

5. (th moment about mean)

-J«-meany f)dt
6. , (th moment about the point x = A)

Median : is kacown as that point which divides the

entire distribution in two
cqual parts Thus, if Mis the median, then
M

Mcan Drviation : Meandeiation abt mean is given by

M)-1
9.
uariles: eisst yuatik (2, san l v siwad ty Molving
 Probability
92

Similartv,Q can be obtaincd by solving

Similarlv. the ith Decile, D, is given by

which fr) is maximum. Thus, it can be obtaine

AL
10. Mode : Mode is the value ofx for
equating the first derivative of f(r) to zero and checking that the secns
by
derivative of f (x) is negative i.e., we use is

f(x) =0and f" (r) <0

provided it lies in [a, b]

2.3.3 Cumulative Distribution Function

As
variables also, there are many
Similar to the discrete case, in case of continuous random
probiems inwhich it is of interest to know the probability that the
value of a continuou
de
random variable X is less than or equal to some real number x.

The cumulative distribution F(r) of continuous random variable X with density f(r) is

Fr) = P(Xs)= |ft) dtfor -o<I<o

The following properties hold good for F(r):

1. 0sF(r)s1,
2. F()= f(x) z0
ie.,F(r) is a non-decreasing function of x

3. F(-«) = lim F(r) = lim

F(u)= lims F()= lim

4. F(x) is a continuous funetion of s on the right,

5 The discontinuities of F(r) are at the
most, countable.
6. PlasX h) = P(a< X<h) = F(b) - F(a) is
7. Since F ()= f (x), we have
 Random
Varlables
and Mathematical Expectation

LFa) =f() ’d F() = f(r) dr

dF(r) iskknown as
probability diflerential of X.
ectrnation 1 4 ; Find a probability density function for the random varlable
whose
strbation
function
is given by
for
F(x) =* for 0 <*<1
for *>1
graph.
Alko draw its
Soltion : Differentiating F(x) for x<0,0 <x<land &>l, as the given density
function
kdiferentiable everywhere except at * =0 and x=l, we get
for t<0
-F(x) =f(r) = for 0 <i<l
for >l
As we know that F(r) is continuous from right and as F(r) is not continuous at x
=0and
r=l, we put f0) and f )both equal to zero as it does not matter how the probability
density is defined at these two points.Thus, we have
for 0 <<l
lo elsewhere
f(X)

0 X

Fig. 2.12

stration 15:Verify that

F(t) = 0 t<0

0st<l/2

=l-3(-), 1/2 st<l

ts a t21 [C.C.S.U. 2002]
distribution function and dertve the density.function ofX.
 9%
Probatu
Solution :It can be seen that

(i) 0sF(t)slfor allt

(ii) F(-o) =0and F{o) =1
(ii) F)is a non-decreasing function and is right continuous for all
Hence, it is a distribution function.
To obtain the density function,we use
d
F(*) =f(1)
Thus, we have

f)=0 I<0
=2x 0sr<l/2
=6.r , 1/2sr<l
=0

The density function is

f() =2r, 0sx<l/2
=6x, l/2 sr <l
=0, otherwise

llustration 16:The probability distribution function of a randomn variable Xis

for 0 <sl
f(r) =2-x for 1sx<2
0 for x 2
Compute the cunulative distribution function of X.
Solution : Here,
For any x in the range x <0, we have

F)- [f9dt - fo.dk -0

For any xin )<x<I

For any xin 1<xs2,

 Varlables and Mathematlcal Expectatlon
Randon

-fod"jvda fe-na

0<r<l
Thus, we have F(r) =
|2r-4-1 lkrs2
x22

24 Joint Probability Distributions

eprevious sections, we have studied random variables related to one-dimensional
espace only. But, there are situations, when we are often interested in recording
* Suitaneous outcomes of several random variables. For example, one may be
value of the
n measuring the temperature (T) and pressure (P) for aconstant
or one may be interestedin knowing the simultaneous heights
(X) and
wins (Y) of girls in a school. Obviously, here we are dealing with a bivariate
two random variables are involved.

foint Probability Mass Function

distribution tor their
two discrete Tandom variables, the probability
by a functionwith values f(r. y) for eah
e vdues(X, y) ) within thebe represented
Tange of Xand Y. This probability distribution is alled as
Se jt
probability
mass function and is given as follows
Jy) =PX =, Y= y) simultaneouslv.
us the probability thatthe outcomes sand yoeur