Welcome to the World of Discrete Distributions!

In this chapter, we are going to explore two of the most important tools in a statistician's toolkit: the Binomial Distribution and the Poisson Distribution. These are "discrete" distributions, which means they deal with things we can count (like the number of heads in a coin flip or the number of emails you receive in an hour).

Don't worry if these terms sound a bit scary at first. By the end of these notes, you'll see that they are just mathematical ways of describing "how likely" something is to happen in the real world. Let’s dive in!

1. The Binomial Distribution (A Quick Refresher)

You might remember the Binomial distribution from your S1 studies. It is used when we have a fixed number of trials and we want to know the number of "successes."

Key Conditions (The BINS Mnemonic)

To use a Binomial model \(X \sim \text{B}(n, p)\), four things must be true:

  • B – Binary: There are only two possible outcomes (Success or Failure).
  • I – Independent: One trial doesn't affect the next one.
  • N – Number: There is a fixed number of trials (\(n\)).
  • S – Same probability: The probability of success (\(p\)) is constant.

Mean and Variance

For the Binomial distribution, you need to remember these two simple formulas (no need to derive them!):

Mean (Expected Value): \(E(X) = np\)
Variance: \(\text{Var}(X) = np(1 - p)\)

Example: If you flip a fair coin 10 times, your expected number of heads is \(10 \times 0.5 = 5\).

Quick Review: The Binomial distribution is for a fixed number of trials where you count successes.

2. The Poisson Distribution

While the Binomial distribution looks at a fixed number of trials, the Poisson Distribution looks at events happening over a fixed interval of time or space.

When do we use Poisson?

Imagine you are standing on a street corner counting how many cars pass by in 10 minutes. Or counting how many chocolate chips are in a cookie. These are Poisson scenarios!

We write this as: \(X \sim \text{Po}(\lambda)\)

Where \(\lambda\) (the Greek letter "lambda") is the average rate of occurrence.

Conditions for a Poisson Model

For a situation to be modeled by a Poisson distribution, events must occur:

  • Independently: One car passing doesn't make another car more or less likely to pass.
  • Singly: Two events cannot happen at exactly the same instant.
  • At a constant average rate: The average number of events per minute stays the same throughout the interval.
  • Randomly: You can't predict exactly when the next event will happen.

The Poisson Formula

To find the probability of exactly \(x\) events happening:

\(P(X = x) = \frac{e^{-\lambda} \lambda^x}{x!}\)

Note: \(e\) is a constant approximately equal to 2.718, and \(x!\) is "x factorial."

Mean and Variance

Here is a neat trick for the Poisson distribution that makes it very easy to remember:

Mean: \(E(X) = \lambda\)
Variance: \(\text{Var}(X) = \lambda\)

In a Poisson distribution, the mean and the variance are equal! If a question shows you data where the mean and variance are very different, a Poisson model might not be appropriate.

Did you know?

The Poisson distribution was named after the French mathematician Siméon Denis Poisson. Interestingly, "Poisson" is also the French word for "fish"!

Key Takeaway: Use Poisson for events happening at a constant rate over time or space. The mean and variance both equal \(\lambda\).

3. The Additive Property of Poisson

One of the most useful things about the Poisson distribution is that \(\lambda\) scales perfectly with the size of the interval.

Scaling the Interval

If the number of emails you get follows \(X \sim \text{Po}(2)\) per hour, then:

  • In 2 hours, the distribution is \(\text{Po}(2 \times 2) = \text{Po}(4)\).
  • In 30 minutes (half an hour), the distribution is \(\text{Po}(2 \times 0.5) = \text{Po}(1)\).

Adding Independent Variables

If you have two independent Poisson variables, \(X \sim \text{Po}(\lambda)\) and \(Y \sim \text{Po}(\mu)\), their sum is also Poisson:

\(X + Y \sim \text{Po}(\lambda + \mu)\)

Quick Tip: Always make sure your \(\lambda\) matches the timeframe mentioned in the probability question!

4. Poisson as an Approximation to Binomial

Sometimes, calculating a Binomial probability is really hard because \(n\) is huge (like 1,000) and \(p\) is tiny (like 0.001). In these cases, we can use the Poisson distribution as a shortcut.

When can we approximate?

You can use \(X \sim \text{Po}(np)\) to approximate \(X \sim \text{B}(n, p)\) if:

  • \(n\) is large (usually \(n > 50\))
  • \(p\) is small (usually \(p < 0.1\))

The new mean \(\lambda\) is simply \(n \times p\).

Analogy: Imagine trying to count how many people in a huge stadium have the exact same birthday as you. The number of people (\(n\)) is huge, but the chance for each person (\(p\)) is small. A Poisson model would fit this perfectly!

Common Mistake: Students often forget to check if \(p\) is small enough. If \(p\) is close to 0.5, you should use the Normal distribution (which you will learn later) instead of Poisson.

5. Working with Cumulative Probabilities

In your exam, you'll often be asked for "at most" or "more than" a certain number. You can use Statistical Tables provided in the exam to save time.

Step-by-Step for "At Least" Questions:

If you need to find \(P(X \geq 3)\) for a Poisson distribution:

  1. Remember that the total probability is always 1.
  2. Identify the opposite of "at least 3," which is "2 or fewer."
  3. Use the formula: \(P(X \geq 3) = 1 - P(X \leq 2)\).
  4. Look up \(P(X \leq 2)\) in the Poisson tables for your specific \(\lambda\).

Encouragement: If the tables don't have your exact \(\lambda\), or if you are asked for a specific value like \(P(X = 4)\), it's often safer and easier to use the formula or your calculator's distribution function!

Summary: Choosing the Right Model

To wrap up, ask yourself these questions when faced with a problem:

  • Is there a fixed number of attempts? Yes \(\rightarrow\) Binomial.
  • Is there a constant rate over time/space? Yes \(\rightarrow\) Poisson.
  • Is \(n\) massive and \(p\) tiny? Yes \(\rightarrow\) Poisson Approximation.

Key Takeaway: Statistics is all about picking the right "shape" to describe the data. Master the conditions for Binomial and Poisson, and you've won half the battle!