RANDOM VARIABLES 1

You aren’t your past.

You are probability of
your future.
Oprah Winfrey

2
3
CONTENTS:

 Properties of Distribution Function

 Density Function

4
PROPERTIES OF CDF

5
Recall : CDF Definition :

The probability that a RV 𝑋 takes a

value less than or equal to 𝑏,
𝑭𝑿 (𝒃) = 𝑷(𝑿 ≤ 𝒃), ∀𝒃
is called the Cumulative Distribution
Function [3].
6
CDF – Property No. 1 (P.1):

𝑭𝑿 𝒙 = 𝑷 𝑿 ≤ 𝒙 , ∀𝒙 ⋯ (2.10)

7
CDF – Property No. 2 (P.2):

𝑭𝑿 𝒙 = 𝑷 𝑿 ≤ 𝒙 , ∀𝒙 ⋯ (2.10)

8
CDF – Property No. 3 (P.3):

𝑭𝑿 𝒙 = 𝑷 𝑿 ≤ 𝒙 , ∀𝒙 ⋯ (2.10)

9
CDF – Property No. 4 (P.4):

At the point of discontinuity 𝒙𝟎 :

+
𝑭𝑿 𝒙𝟎 = 𝑭𝑿 𝒙𝟎
−
𝑭𝑿 𝒙𝟎 ≠ 𝑭𝑿 𝒙𝟎
10
CDF – Properties :

The properties (2.41)–(2.44), altogether,

may be used TO TEST IF a given function
could be a valid distribution function.

11
CDF – Properties :

12
CDF – Properties :

13
CDF – Properties :

14
WHAT CDF IS USED FOR?

15
CDF – Computing Probability of an Event :

As 𝑋 ≤ 𝑥2 = {−∞ < 𝑋 ≤ 𝑥2 } & 𝑋 ≤ 𝑥1 = {−∞ < 𝑋 ≤ 𝑥1 }

Then 𝑋 ≤ 𝑥2 − 𝑋 ≤ 𝑥1 = {𝑥1 < 𝑋 ≤ 𝑥2 }
16
CDF – Computing Probability of an Event :
𝑷 𝒙𝟏 < 𝑿 ≤ 𝒙𝟐 = 𝑷 𝑿 ≤ 𝒙𝟐 − 𝑷 𝑿 ≤ 𝒙𝟏 ⋯ (𝟐. 𝟒𝟓)

Left-Open interval.

17
CDF – Computing Probability of an Event :

𝑷 𝒙𝟏 < 𝑿 ≤ 𝒙𝟐
= 𝑷 𝑿 ≤ 𝒙𝟐 − 𝑷 𝑿 ≤ 𝒙𝟏 ⋯ (𝟐. 𝟒𝟓)
= 𝑭𝑿 𝒙𝟐 − 𝑭𝑿 𝒙𝟏 ⋯ (𝟐. 𝟒𝟔)

18
EX 2.2.5

19
CDF – Computing Probability of an Event :

20
CDF – Computing Probability of an Event :

21
CDF – Computing Probability of an Event :

22
CDF – Computing Probability of an Event :

23
CDF – Computing Probability of an Event :

24
EX 2.2.6

25
CDF – Computing Probability of an Event :

26
CDF – Computing Probability of an Event :

𝑥−2
𝐹𝑋 (𝑥) =
4

For continuous RVs, we don’t assign

probabilities to point events rather
we assign probabilities to intervals. 27
 DENSITY
FUNCTION
28
Probability Density Function – PDF :
o The distribution function 𝑭𝑿 (𝒙) introduced in
Sect. 2.2 is a mathematical tool used to describe
a RV.
o However, it is often cumbersome to work with
distribution especially, when dealing with the
continuous RVs.
o The easiest alternative is to use the probability
density function (PDF).
29
PDF – Definition :

30
PDF – Example :

31
RECALL :
 The unit step and unit impulse are closely related.
 In DT the unit impulse is the first difference of the unit step,
and the unit step is the running sum of the unit impulse.
 Correspondingly, in CT the unit impulse is the derivative of
the unit step, and the unit step is the running integral of the
impulse.
 NB: The fact that it is a first difference and a running sum that relate the step and the
impulse in DT and a derivative and running integral that relate them in CT should not be
misinterpreted to mean that a first difference is a good "representation" of a derivative or
that a running sum is a good "representation" of a running integral. Rather, for this particular
situation those operations play corresponding roles in continuous time and in discrete time.
32
RECALL :

𝛿 𝑛 = 𝑢 𝑛 − 𝑢[𝑛 − 1]
∞

𝑢 𝑛 = ෍ 𝛿[𝑛 − 𝑘]
𝑘=0
33
CDF – PDF Relationship :

34
CDF – PDF Relationship :

𝒙𝟎

𝑭𝒙 𝒙𝟎 = න 𝒇𝑿 𝜶 𝒅𝜶
−∞
35
CDF – PDF Relationship :

𝑥0

𝐹𝑋 𝑥0 = න 𝑓𝑋 𝑣 𝑑𝑣
−∞

𝑭𝑿 𝒙 = 𝑷 𝑿 ≤ 𝒙 , ∀𝒙 ⋯ (2.10) 36
CDF – PDF Relationship [4]:
𝑥

𝐹𝑋 𝑥 = 𝑃 𝑋 ≤ 𝑥 = න 𝑓𝑋 𝑡 𝑑𝑡
−∞

37
PDF – Explanation : 𝑷 𝒙𝟏 < 𝑿 ≤ 𝒙𝟐 = 𝑭𝑿 𝒙𝟐 − 𝑭𝑿 𝒙𝟏 ⋯ (𝟐. 𝟒𝟔)

38
PDF – Explanation :

This suggests us that

probability of an
interval is equal to
area under the PDF
curve over that
interval.

39
PDF – Explanation :

One Strip = Infinitesimal Interval

𝑃 𝑥 < 𝑋 ≤ 𝑥 + Δ𝑥 = 𝑓𝑋 𝑥 𝑑𝑥
Interval of Arbitrary size
𝑏

𝑃 𝑎 < 𝑋 ≤ 𝑏 = න 𝑓𝑋 𝑥 𝑑𝑥
𝑎

40
MATLAB SIMULATIONS

41
PDF – Explanation :

More about it!

See Sec. 2.4

42
EX 2.3.1

43
PDF – Example 2.3.1:

44
PDF – Example 2.3.1:

45
PDF – Example 2.3.1:

46
PDF – Example 2.3.1:

47
DELTA FUNCTION

48
DELTA Function :

49
DELTA Function :

50
DELTA Function :

51
DELTA Function :

52
DELTA Function :

53
DELTA Function :

54
DELTA Function :

55
PDF OF DISCRETE
RVS ( a.k.a. PMF)
56
RECALL :

The array of probabilities

𝑃{𝑋 = 𝑥𝑘 } is called the
probability mass function
(PMF).
57
RECALL :

58
PDF of a Discrete RV :

59
PDF of a Discrete RV :

60
EX 2.3.2

61
PDF of a Discrete RV – Example 2.3.2 :

1 1 1
𝐹𝑋 𝑥 = 𝑢 𝑥 − 1 + 𝑢 𝑥 − 2 + ⋯ + 𝑢(𝑥 − 6)
6 6 6

62
PDF of a Discrete RV – Example 2.3.2 :

1 1 1
𝐹𝑋 𝑥 = 𝛿 𝑥 − 1 + 𝛿 𝑥 − 2 + ⋯ + 𝛿(𝑥 − 6)
6 6 6

63
PDF of a Discrete RV – Example 2.3.2 :

𝐹𝑋 3 = න 𝑓𝑋 𝑣 𝑑𝑣
−∞

64
 PDF OF
MIXED RVS
65
PDF of a Mixed :

66
EX 2.3.3

67
PDF of a Mixed – Example :

68
PDF of a Mixed – Example :

69
PDF of a Mixed – Example : SLOPE = 𝒅𝒚 / 𝒅𝒙
𝟐 𝟏 𝟏
= × =
𝟑 𝟒 𝟔

70
PDF of a Mixed – Example :

71
PROPERTIES OF PDF

72
PDF – Property No. 1 (P.1):

73
PDF – Property No. 2 (P.2):

74
PDF – Property No. 3 (P.3):

Left-Open interval.

75
PDF – Property
No. 3 (P.3):

76
PDF – Property No. 3 [4]:

77
PDF – Property No. 1 & 2 (P.1 + P.2):

The properties (2.82) and (2.83) serve as a

TEST to determine whether a given
function is a PDF.
Both properties must be satisfied in order
to determine that a function is a PDF.
78
EX 2.3.4

79
PDF Properties – How to Apply :

(2.87) is a special case of Laplace Distribution with 𝜇 = 0.

80
PDF Properties – Example 2.3.4 :

AuC = 1.3333
81
PDF Properties – Example 2.3.4 :

AuC = 1.0000

82
PDF Properties – Example 2.3.4 :

83
Example 2.3.4 :
 Find the probability that the RV X is
negative, if a=1.
 Find the probability that the RV X is in the
interval [1, 2] , if a=1.

84
Example 2.3.4 :

The probability is
presented as a
SHADED AREA
in the PDF, and
as a POINT in
the distribution.
85
Example 2.3.4 : The probability is
presented as a
SHADED AREA
in the PDF.

86
Example 2.3.4 :

The probability is
presented as a
difference of
Two POINTs in
the distribution.

87
DISCLAIMER
88
These power point slides are NOT
SUBSTITUTE of reading TEXT
BOOK(S).
You’re ALWAYS DIRECTED to
CAREFULLY READ the relevant
book chapter and SOLVE ALL
examples and end problems.
89
REFERENCES :
[1] [Dolecek-2013]

[2] [Kay-2005]

[3] [Sugiyama-2016] Introduction to Statistical Machine Learning

(Annotated)

[4] [Gubner-2006] Probability and Random Processes for

Electrical and Computer Engineers

90