A number assigned to each possible outcome of something random. Ex: X = number of heads in 200 coin flips X=148 is an event, true or false P(X=148) is a probability, a number between 0 and 1.
Probability mass function (PMF) A function that assigns a probability to each possible value of a discrete random variable, the chance of each value.
Defined by:
It must satisfy two properties:
- Non-negativity: for every .
- Normalization: .

Expectation (also called the Mean), is the balance point of the PMF, defined as:
For a fair 6-sided die, each outcome has an equal probability of .


Types of Random Variables (Distribution)
Bernoulli Random Variable
A random variable with two possible values (0 or 1).
is a Bernoulli random variable:
Expectation
Examples
- Coin flip
- Random binary digit (bit)
Binomial Random Variable
Represents independent trials of a Bernoulli random variable.
- is the number of successes (1’s) in trials.
is a Binomial random variable:
Expectation
Examples
- Number of heads in coin flips.
- Number of 1’s in a randomly generated string of bits.
Poisson Random Variable
Approximates a Binomial distribution when is large and is small.
- takes on values
is a Poisson random variable:
Expectation
When we approximate a Binomial distribution with a Poisson distribution, we choose:
Generally Good Approximation When
- and
- and
Examples
- Number of component failures in a large data center.
- Number of bits corrupted in message transmission.
Poisson Distribution The Poisson distribution models the number of events occurring within a fixed interval of time or space.
The PMF is:
Can use Poisson distribution for
- Temporal Modelling (Number of events per fixed interval of time)
Definition: Counting the number of events occurring within a fixed interval of time.
Example: Counting Flashes in Radioactive Decay. Researchers counted the flashes produced by radioactive decay in each 7.5-second window.
The Poisson model
- Divide the 7.5-second window into many extremely small time intervals (slices).
- Each slice has a very small probability of containing a flash.
- The number of slices is very large.
- Count the total number of flashes occurring during the entire 7.5-second window.
- is large.
- is small.
The total number of flashes can be approximated by a Poisson random variable.
If the average rate of flashes per second is , then:

- Spatial Modelling (Number of events per fixed amount of space)
Definition: Counting the number of events occurring within a fixed amount of space.
Example Suppose we are:
- Making many cookies with lots of chocolate chips.
- Dividing the batter into equally sized portions.
- Distributing the chocolate chips randomly.
We want to count the number of chocolate chips in one particular cookie.
The Poisson model
- Each chocolate chip has a small probability of ending up in our chosen cookie.
- There are many chocolate chips.
- Each chip can be treated as a Bernoulli trial.
The probability that a particular chip ends up in a specific cookie is:
The number of chocolate chips in a particular cookie follows a Binomial distribution:
When is large and is small, we can approximate it using a Poisson distribution:
Where:
If there are 1,000 chocolate chips and 100 cookies:
The expected number of chocolate chips per cookie is 10.
A Poisson model is appropriate when events occur independently at an approximately constant average rate or density.
Variance
Intuitively, variance measures the spread, how far values wander from the mean.
- Often use standard deviation to refer to the “spread”.
All three are discrete random variables, but their possible values differ
Bernoulli
- or
- Only two possible outcomes.
Binomial
- Values range from 0 to .
Poisson
- Non-negative integers with no upper limit.
Continuous Random Variables

For continuous random variables, the probability of any exact value is zero.
- We calculate probabilities over ranges:
- Probability Density Functions (PDFs) are shapes showing how likely different possible outcomes are.
- Continuous RVs use a Probability Density Function (PDF).
- Probability is the area under the PDF curve between and .
Uniform(0,1) Random Variable
A Uniform distribution has a flat probability density, meaning equal-length intervals are equally likely.
- The probability density is 1, but that doesn’t mean every outcome has a probability of 1.

- takes values between and .
- Probability depends on the length of the interval.
For :
Example: A bus arrives at a random time between 0 and 10 minutes, with every moment equally likely.
Exponential Random Variable
- Poisson: X = “how many events in this time window?”
- Exponential: X = “how long until the next event?”
Exponential is going to give you time outcomes, Models time

The exponential random variable is written as:
- It uses the same parameter as the Poisson distribution.
- can take values from to .
- Smaller values are more likely than larger values.
- As increases, the probability density decreases.
The graph starts high near and decreases as increases.
Exponential distributions model waiting times, where short waits are common and long waits are rare.

Cumulative Distribution Function (CDF)
A Cumulative Distribution Function (CDF) is a shortcut for calculating the probability (area) under a PDF up to a given value.
- The CDF gives the probability that is less than or equal to .
- It measures the area under the PDF from the beginning up to .

- PDF: Shows probability density.
- CDF: Gives cumulative probability up to .
- The CDF makes it easy to calculate probabilities without manually finding the area under the PDF.
_Exponential Distribution Example Let be the years until the next big (magnitude 8+) earthquake.

The probability that the next big earthquake happens within 1 year.
We get this by plugging into the CDF
Normal Random Variable
A continuous random variable with a bell-shaped curve
Two parameters
-
(mu) = Mean (average).
-
= Variance.
-
Values closer to the mean () have higher probability density.
-
can take any value from to .
