§3.1 -3.

5 Discrete
Random Variables

Dr. Tom Reid

[email protected] or 803-641-3536
Office: PEN 237-L
For next time
Objectives
 Know basic concepts of discrete random
variables (RVs)
 Distributions: pmf vs CDF
 Measures of location: Mean, median
 Measures of spread: Variance, standard deviation
Discrete Random Variables
 Discrete RVs:
 can take on only a countable (but possibly infinite)
number of values
 possible values form a discrete set
 Examples
 S={0,1}
 S={0,1,2,3,…}
 S={-1,5,10}
Discrete Distributions
 Discrete Probability Distribution: a rule (table, equation)
that associates a probability to every possible value of a
discrete RV
 No negative probabilities
 Must sum to 1
 Sometimes called “probability mass function” (pmf):
p(x) = P(X = x)

Ex: are these distributions?
 x: 1 2 3 4 5 Yes
P(X=x) .3 .3 .1 .1 .2
 x: 1 2 3 4 5 No (sum > 1)
P(X=x) .2 .2 .3 .2 .2
 x: 1 2 3 4 5 No (neg. prob.)
P(X=x) .3 .3 .3 .3 -.2
Example: Three Relays

 Oi: relay i is open (Ci for closed), i = 1, 2, 3

 S = {O1∩ O2∩ O3, O1∩ O2∩ C3 ,
O1∩ C2∩ O3, O1∩ C2∩ C3 ,
C1∩ O2∩ O3, C1∩ O2∩ C3 ,
C1∩ C2∩ O3, C1∩ C2∩ C3 } x P(X=x)
 P(Oi) = ¼ 0 27/64
 Let X = number of open relays 1 27/64
 Find P(X=x) for x=0,1,2,3 2 9/64
 Note: requires assumption of 3 1/64
independence!
Relay Example (cont) x
0
P(X=x)
27/64
1 27/64
2 9/64
 What is the chance that a circuit with these 3 1/64
relays in series will work?
 All must be closed to work, so…
 P(series circuit works) = P(X = 0) = 27/64
 What is the chance that a circuit with these
relays in parallel will work?
 At least one must be closed, so…
 P(parallel circuit works) = P(X < 3)
= P(X = 0) + P(X = 1) + P(X = 2)
= 63/64
 Easier: use the complement!
Cumulative Distribution Function

 The probability mass function (pmf) specifies

the probability that a random variable is equal to
a given value
 p(x) = P(X = x)
 A function called the cumulative distribution
function (CDF) specifies the probability that a
random variable is less than or equal to a given
value
 The cumulative distribution function of the random
variable X is the function F(x) = P(X ≤ x).

7
Relay Example (redux)
 Recall the example with three relays?
 X counts number of open relays
 P(Oi) = ¼
 Compute the CDF of the random variable X
x P(X=x) P(X<x)
0 27/64 27/64
1 27/64 54/64
2 9/64 63/64
3 1/64 64/64 = 1
Discrete Random Variable
(recap)
Let X be a discrete random variable. Then
 The probability mass function of X is the function
p(x) = P(X = x).

 The cumulative distribution function of X is the

function F(x) = P(X ≤ x).
 F ( x)  p (t )  P ( X t )
t x t x

  p( x)  P( X x) 1 , where the sum is over all

x x
possible values of X.

9
Mean of a Random Variable
 Mean (“expected value”):
k
E ( X )   X  x1 p1  x 2 p2  x k p k   xi pi
i 1
 Note: E(X) is a “linear operator”
 E(aX+b) = a E(X) + b x P(X=x)
 E(X+Y) = E(X) + E(Y) 0 27/64
 Find E(X) for the Relay Example 1 27/64
 μ = 0(27/64) + 1(27/64) + 2(9/64) + 3(1/64) 2 9/64
X
= 48/64 = 3/4 3 1/64
 Interpretation: On average,

if we repeat this experiment a million times

and averaged the number of open circuits seen,
the average number of open circuits would be 0.75
 Mean, Variance and Std.
Deviation of a Random Variable
 Variation: Var ( X )  X2
Careful! This is not
at all obvious!! 
E  X   
2
   x   
i
2
pi

 E  X    E X    xi2 pi   2
2
2
 
 Standard Deviation: SD( X )  X   2
X

 Note: Units for variation is the square of the units for X

Units for standard deviation is the same as for X
 For Relay Example:
 σ2X = (0-3/4)2(27/64) + (1-3/4)2(27/64)
+ (2-3/4)2(9/64) + (3-3/4)2(1/64) = .5625
 σX = sqrt(.5625) = .75
Example: Profits and
Weather
Schedule a concert to maximize expected profit. Table gives profit in $1000.
 WeatherProfit Indoors Profit Outdoors Probability
 Sunny and Warm 50 80 0.5
 Sunny and Cold 50 60 0.2
 Rainy and Warm 60 40 0.2
 Rainy and Cold 40 6 0.1

 Expected Indoors Profit? (call that “IP”)


μIP = E(IP) = 50(.5) + 50(.2) + 60(.2) + 40(.1) = 51

σ2IP = Var(IP) = (50-51)2(.5) + (50-51)2(.2) + (60-51)2(.2) + (40-51)2(.1)
= 29

σIP = StDev(IP) = sqrt(29) = 5.385
 Expected Outdoors Profit? (“OP”)

μOP = E(OP) = 80(.5) + 60(.2) + 40(.2) + 6(.1) = 60.6

σ2OP = Var(OP)
= (80-60.6)2(.5) + (60-60.6)2(.2) + (40-60.6)2(.2) + (6-
60.6)2(.1)
= 571.24

σOP = StDev(OP) = sqrt(571.24) = 23.9
Functions of a Single RV
 If X has pmf f(x), and h( ) is a function then
E  h( X )   h( x) f ( x)


x
 In particular, if h(x) = ax+b, where a and b are
constants, then
 aX b a X  b ,

 aX
2
b a X
2 2 ,

  aX b  a  X .

13
Linear Functions of Multiple
RVs
If X and Y are random variables, and a and b
are constants, then
aX bY aX  bY a X  bY .

More generally, if X1, …, Xn are random

variables and c1, …, cn are constants, then the
mean of the linear combination c1 X1+…+cn Xn is
given by
c X c X ...c X c1 X  c2  X  ...  cn  X .
1 1 2 2 n n 1 2 n

14
Two Independent Random
Variables
If X and Y are independent random variables,
and S and T are sets of numbers, then
P ( X  S and Y  T ) P ( X  S ) P (Y  T ).

More generally, if X1, …, Xn are independent

random variables, and S1, …, Sn are sets, then
P( X 1  S1 , X 2  S2 ,K , X n  Sn )
P ( X 1  S1 ) P ( X 2  S2 )L P ( X n  S n ) .

15
Variance Properties
If X1, …, Xn are independent random variables,
then the variance of the sum X1+ …+ Xn is
given
by  X2  X ... X  X2   X2  ....   X2 .
1 2 n 1 2 n

If X1, …, Xn are independent random variables

and c1, …, cn are constants, then the variance
of the linear combination c1 X1+ …+ cn Xn is
given2 by
 c X c X ...c X c12 X2  c22 X2  ....  cn2 X2 .
1 1 2 2 n n 1 2 n

16
More Variance Properties

If X and Y are independent random variables

with variances X 2
and  2
Y , then the variance of

the sum X + Y is
 2
X Y    .
2
X
2
Y

The variance of the difference X – Y is

 2
XY    .
2
X
2
Y

17
For Next Time
 Work problems from WileyPLUS
 Read §3.5-3.9