Welcome to the Poisson Distribution!

In this chapter, we are going to learn how to predict the "unpredictable." Have you ever wondered how a call center knows how many staff to hire for a Tuesday afternoon, or how a hospital estimates the number of patients arriving at the emergency room? They use the Poisson Distribution!

We use this distribution to model the number of times an event occurs within a fixed interval of time or space. Don't worry if this seems a bit abstract right now—we will break it down step-by-step!

1. What Makes a Poisson Distribution?

Before we use the math, we need to know when it is appropriate to use this model. For a situation to follow a Poisson distribution, the events must happen:

  • Independently: One event happening doesn't change the chance of another happening. (If one person enters a shop, it doesn't "cause" another person to enter).
  • Randomly: Events can happen at any time.
  • Singly: Two events cannot happen at the exact same heartbeat of time.
  • At a Constant Average Rate: The average number of events (which we call \(\lambda\), or "lambda") stays the same throughout the interval.

Quick Analogy: Imagine raindrops landing on a specific square tile during a steady drizzle. The drops land randomly, independently, and they usually don't land at the exact same spot at the exact same time. The average number of drops per minute remains constant.

Key Takeaway:

Memory Aid: Think of the word CRIS: Constant rate, Random, Independent, Singly. If these four boxes are ticked, you have a Poisson variable!

2. The Poisson Formula

The Poisson distribution is written as:
\(X \sim Po(\lambda)\)

Here, \(X\) is our random variable (the number of successes), and \(\lambda\) (lambda) is the mean number of occurrences in the given interval.

To find the probability of exactly \(r\) successes, we use this formula:
\(P(X = r) = \frac{e^{-\lambda} \lambda^r}{r!}\)

Breaking down the symbols:

  • \(e\): A mathematical constant approximately equal to 2.718 (you have a button for this on your calculator!).
  • \(\lambda\): The average number of events.
  • \(r\): The number of events you are interested in (e.g., "What is the probability of 3 cars passing?").
  • \(r!\): "r factorial" (e.g., \(3! = 3 \times 2 \times 1\)).

Step-by-Step Example:
The average number of emails a student receives per hour is 4. Find the probability they receive exactly 2 emails in the next hour.
1. Identify \(\lambda\): Here, \(\lambda = 4\).
2. Identify \(r\): We want \(r = 2\).
3. Plug into the formula: \(P(X=2) = \frac{e^{-4} \times 4^2}{2!}\).
4. Calculate: \(P(X=2) = \frac{0.0183 \times 16}{2} = 0.1465\) (to 4 decimal places).

Quick Review:

Always check your interval! If the average is 4 per hour, but the question asks for the probability in two hours, you must double your \(\lambda\) to 8.

3. Mean and Variance: The Identical Twins

One of the coolest (and easiest!) things about the Poisson distribution is its mean and variance.

For a Poisson distribution \(X \sim Po(\lambda)\):
Mean \(E(X) = \lambda\)
Variance \(Var(X) = \lambda\)

Did you know? If a question tells you the mean and variance of a dataset are roughly equal, it is a very strong hint that the data follows a Poisson distribution!

4. Adding Independent Poisson Variables

Sometimes we have two different Poisson events happening at once. If \(X\) and \(Y\) are independent Poisson variables:

If \(X \sim Po(\lambda_a)\) and \(Y \sim Po(\lambda_b)\),
Then \(X + Y \sim Po(\lambda_a + \lambda_b)\).

Example: If a bakery sells an average of 3 white loaves per hour and 2 brown loaves per hour, the total number of loaves sold follows \(Po(3 + 2) = Po(5)\).

5. Approximating the Binomial Distribution

Sometimes, a Binomial distribution (where you have a fixed number of trials \(n\)) is so large that it is easier to use Poisson. We can do this when:

  1. \(n\) is large (usually \(n > 50\)).
  2. \(p\) is small (the probability is low, usually \(np < 5\)).

In this case, we use \(\lambda = np\).

Why do this? Calculating \(^{100}C_{5} \times (0.01)^5 \times (0.99)^{95}\) is much harder than using the Poisson formula. It's a mathematical shortcut!

Common Mistake to Avoid:

Don't use the Poisson approximation if \(p\) is large (close to 0.5). That's when we use the Normal distribution instead!

6. Approximating Poisson with the Normal Distribution

Wait, we can turn Poisson into a Normal distribution too? Yes! If \(\lambda\) is large (usually \(\lambda > 15\)), the Poisson graph starts to look like a symmetrical bell curve.

In this case, we use:
\(X \sim N(\lambda, \lambda)\)
(Where the mean is \(\lambda\) and the variance is also \(\lambda\)).

Important: Continuity Correction!
Because Poisson is discrete (you can't have 4.5 emails) and Normal is continuous (you can have 4.5 cm), you must adjust your values by 0.5 when switching between them.
Example: To find \(P(X > 20)\) using the Normal approximation, you would actually calculate \(P(X > 20.5)\).

Key Takeaway:

Check your \(\lambda\):
- If \(\lambda\) is small: Use the standard Poisson formula.
- If \(\lambda\) is large (\(> 15\)): Use the Normal approximation with a continuity correction.

Summary Checklist

Before you sit your exam, make sure you can:

  • List the 4 conditions for a Poisson distribution (CRIS).
  • Calculate probabilities using the \(e^{-\lambda}\) formula.
  • Adjust \(\lambda\) for different time intervals.
  • Identify when to use Poisson to approximate a Binomial.
  • Use the Normal distribution to approximate a Poisson when \(\lambda\) is large (and remember the 0.5 correction!).

You've got this! Poisson might look scary with its Greek letters, but it's just a way to count random events. Keep practicing those past paper questions!