Session 2: Probability distributions

and density functions

Susan Thomas
http://www.igidr.ac.in/susant
[email protected]

IGIDR
Bombay

Session 2: Probability distributionsand density functions p. 1

Recap
The definition and scope of probability in the
domain of an event space.
Basic properties of probability
Notion of mutually exclusive
Notion of conditional and unconditional probability
Notion of independance

Session 2: Probability distributionsand density functions p. 2

The aim of this session
1. Discrete and continuous random variables
2. Probability distributions
3. Case 1: The bernoulli distribution
4. Case 2: The binomial distribution
5. Probability density functions
6. Case 3: The uniform distribution
7. Case 4: The normal distribution
8. Cumulative distributions and density functions
9. Sampling from distributions: population vs. sample

Session 2: Probability distributionsand density functions p. 3

Random variables

Session 2: Probability distributionsand density functions p. 4

Random Variables
A random variable is what an outcome can be. It
could also be a function of an outcome.
For example, the value of the first roll of a die.
For example, the sum of a roll of two dice.
Discrete random variables: where the possible
events are countable.
For example, the roll of a dice, or the outcome of a
horse race, or whether the firm will default or not.
Continuous random variables: where the possible
events are not countable.
For example, the number of white hair on my head,
or how much dividend INFOSYSTCH will
announce next year, or the price of Citibank stock.
Session 2: Probability distributionsand density functions p. 5
Note: Continuous RVs can have a fixed minimum or
Probability distributions

Session 2: Probability distributionsand density functions p. 6

What is a probability distribution?
For a discrete RV, the probability distribution (PD) is
a table of all the events and their related
probabilities.
For example, in the roll of a die:
Value Probability Value Probability

1 1/6 4 1/6
2 1/6 5 1/6
3 1/6 6 1/6

A probability distribution will contain all the

outcomes and their related probabilities, and the
probabilities will sum to 1. Session 2: Probability distributionsand density functions p. 7
How to read a probability distribu-
tion
From the distribution, we can find:
X=3
Pr(X = 3) = 1/6
X = even number
Pr(X = 2 or X = 4 or X = 6) = 3/6 = 1/2
More interesting, we can also find:
Pr(X > 3) = 3/6 = 1/2
This is called a cumulative probability.

Session 2: Probability distributionsand density functions p. 8

What is a cumulative probability dis-
tribution (CD)?
A table of the probabilities cumulated over the
events.

Value Probability Value Probability

X1 1/6 X4 4/6
X2 2/6 X5 5/6
X3 3/6 X6 1

The CD is a monotonically increasing set of

numbers
The CD always ends with at the highest value of 1.
Session 2: Probability distributionsand density functions p. 9
Examples of a PD: Bernoulli RVs
Bernoulli distribution: The outcome is either a failure
(0) or a success (1).
X is a bernoulli RV when
Pr(X = 0) = p
Pr(X = 1) = (1 p)

For example, the USD-INR rises or not at the end of

the day.
The bernoulli distribution has one parameter, p.

Session 2: Probability distributionsand density functions p. 10

The CD of a Bernoulli RV
We need to know what p of the RV is. Say p = 0.35.
The CD of this Bernoulli RV is:
Value Probability

X0 0.35
X1 1.00

Session 2: Probability distributionsand density functions p. 11

Examples of a PD: Binomial RVs
Binomial distribution: The outcome is a sum (s) of a set
of bernoulli outcomes (n).
For example, the number of times USD-INR rose in
the last 10 days?

n!
Pr(X = s) = ps (1 p)(ns)
(n s)!s!
Here, p is the probability that the USD-INR rose in a
day. The assumption is that p is constant over these
10 days.
The binomial distribution has two parameters: p, n.
Session 2: Probability distributionsand density functions p. 12
Explaining n!
n! is mathematical short-hand for the product
1 2 3 4 . . . (n 2) (n 1) n
Example: 5! = 1 2 3 4 5
Note: 0! = 1! = 1

Session 2: Probability distributionsand density functions p. 13

Testing concepts: Binomial RV CD
A bernoulli event has a probability of 0.4. What is the
CD of a binomial RV that is the number of times that
success can be acheived in 4 trials?

Session 2: Probability distributionsand density functions p. 14

Testing concepts: Binomial RV CD
A bernoulli event has a probability of 0.4. What is the
CD of a binomial RV that is the number of times that
success can be acheived in 4 trials?
Value PD CD

0 0.1296 0.1296
1 0.3456 0.4752
2 0.3456 0.8208
3 0.1536 0.9744
4 0.0256 1.0000

Session 2: Probability distributionsand density functions p. 14

Testing concepts: A discrete RV
A discrete RV has the following PD:

X 1 2 4 5 8

Pr(X) 0.20 0.25 0.30 0.10

1. Find Pr(4).
2. Find ((Pr(x) = 2) or (Pr(x) = 4)).
3. Find Pr(x 4).
4. Find Pr(x < 4).

Session 2: Probability distributionsand density functions p. 15

Testing concepts: A discrete RV
A discrete RV has the following PD:

X 1 2 4 5 8

Pr(X) 0.20 0.25 0.30 0.10

1. Find Pr(4). 0.15

2. Find ((Pr(x) = 2) or (Pr(x) = 4)). 0.55
3. Find Pr(x 4). 0.60
4. Find Pr(x < 4). 0.45

Session 2: Probability distributionsand density functions p. 15

Probability density functions

Session 2: Probability distributionsand density functions p. 16

What is a probability density func-
tion?
The probability density function (PDF) is the PD of
a continuous random variable.
Since continuous random variables are uncountable,
it is difficult to write down the probabilities of all
possible events.
Therefore, the PDF is always a function which gives
the probability of one event, x.
If we denote the PDF as function f , then
Pr(X = x) = f (x)
A probability distribution will contain all the
outcomes and their related probabilities, and the
probabilities will sum to 1. Session 2: Probability distributionsand density functions p. 17
The problem with estimating PDFs
In a set of continuous random variables, the
probability of picking out a value of exactly x is
zero.
We define the Pr(X = x) as the following difference:

Pr(X (x + )) Pr(X x)
as becomes an infinitesimally small number.
Here, Pr(X x) is the cumulative density function
of X.

Session 2: Probability distributionsand density functions p. 18

What is the cumulative density func-
tion (CDF)?
Analogous to the discrete RV case, the CDF is the
cumulation of the probability of all the outcomes
upto a given value.
Or, the CDF is the probability that the RV can take
any value less than or equal to X.
If we assume that the RV X can take values from
to , then theoretically,
Z X
F (X) = f (x)d(x)

Session 2: Probability distributionsand density functions p. 19

Reformulating the PDF in calculus
Pr(x = X) is given as:

Pr(x = X) = F (X + ) F (X)
= d(F (x))/d(x)

for infinitesimally small .

For continuous RVs, we approach the Pr(x) as the
derivative of the CDF.

Session 2: Probability distributionsand density functions p. 20

Examples of a PDF: Uniform RVs
Uniform distribution: The outcome is any number that
can take a value between a minimum (A) and a
maximum (B) with equal probability.
For a uniform RV,
Pr(X = x) = 1/(B A)
The uniform density has two parameters, A, B.

Session 2: Probability distributionsand density functions p. 21

Testing concepts: Uniform density
CD
What is the form of the Uniform CDF, given that the
maximum = B and minimum = A?

Session 2: Probability distributionsand density functions p. 22

Testing concepts: Uniform density
CD
What is the form of the Uniform CDF, given that the
maximum = B and minimum = A?
Z X
F (X) = f (x)d(x)
minimum
Z X
= 1/(B A)d(x)
A
= (X A)/(B A)

Session 2: Probability distributionsand density functions p. 22

Examples of a PDF: Nor-
mal/Gaussian RV
1 1 2
((x)/)
f (x) =
2
e 2

The normal has two parameters, , .

A lot of economic variables are assumed to be
normal distributed.

Session 2: Probability distributionsand density functions p. 23

Features of the normal PDF
RVs can take values from to .
It is symmetric: Pr(x) = Pr(x)
When X is a normal RV with parameters , , then
Y = 5.6 + 0.2X will also be a normal RV, with
known parameters (5.6 + 0.2), (0.2).
Note: Special case of a normal distribution is
= 0, = 1. This is called a standard normal
distribution.

Session 2: Probability distributionsand density functions p. 24

Probability distributions
1.0

Normal density
Normal probability
0.8
0.6
0.4
0.2
0.0

4 2 0 Session22: Probability distributionsand density

4 functions p. 25
Problems to be solved

Session 2: Probability distributionsand density functions p. 26

Q1: Binomial RVs
A door-to-door salesperson has found that her success
rate in selling is 0.2. If the salesperson contacts three
persons, what is the probability that
1. She sells to all three?
2. She sells to exactly one?
3. She sells to at least one?

Session 2: Probability distributionsand density functions p. 27

Q2: Uniform density CD
If B = 10 and A = 5
1. What is Pr(x = 5.5)?
2. What is Pr(x = 4.5)?
3. What is Pr(x > 7)?

Session 2: Probability distributionsand density functions p. 28

Q3:

Session 2: Probability distributionsand density functions p. 29

References
Chapter 2, S HELDON ROSS. Introduction to
Probability Models. Harcourt India Pvt. Ltd., 2001,
7th edition

Session 2: Probability distributionsand density functions p. 30