Welcome to Discrete Probability Distributions!
In this chapter, we are moving beyond basic probability to look at how we can model real-world events using Discrete Probability Distributions. This is a core part of Further Statistics 1. Whether you are counting how many emails you get in an hour or how many attempts it takes to pass a level in a game, these distributions help us predict the "long-run" outcome.
Don't worry if the formulas look a bit intimidating at first! We’ll break them down into simple steps and use analogies to help them stick. Let’s dive in!
1. The Fundamentals: Mean and Variance
Before we look at specific distributions, we need to know how to find the "center" and the "spread" of any discrete random variable \(X\).
The Mean (Expected Value)
The Expected Value, written as \(E(X)\) or \(\mu\), is basically the long-term average. If you performed an experiment thousands of times, what would the average result be?
Formula: \(E(X) = \sum x P(X=x)\)
Think of it like this: Multiply every possible outcome by the chance of it happening, then add them all up.
The Variance
The Variance, written as \(Var(X)\) or \(\sigma^2\), measures how much the results "shake" or spread out from the mean.
Formula: \(Var(X) = E(X^2) - [E(X)]^2\)
Where \(E(X^2) = \sum x^2 P(X=x)\).
Extending to Functions: \(E(g(X))\)
Sometimes we don't just want the mean of \(X\), but the mean of a function of \(X\), like \(X^2\) or \(3X+2\). This is written as \(E(g(X))\).
The Trick: Just replace the \(x\) in your sum with the function \(g(x)\).
Formula: \(E(g(X)) = \sum g(x) P(X=x)\)
Quick Review Box:
1. \(E(X)\) is the "Average result".
2. \(Var(X)\) is the "Spread". Always remember: "Mean of the squares minus the square of the mean."
2. The Poisson Distribution
The Poisson Distribution is perfect for modeling random events that happen at a constant average rate over a specific interval of time or space.
Real-world examples: The number of cars passing a point in 10 minutes, or the number of typos on a page of a book.
Conditions for Poisson
For a situation to be modeled by \(X \sim Po(\lambda)\), events must occur:
1. Independently (one event doesn't affect the next).
2. Singly (two events can't happen at the exact same instant).
3. At a constant average rate (\(\lambda\)).
Key Properties
If \(X \sim Po(\lambda)\):
- Mean: \(E(X) = \lambda\)
- Variance: \(Var(X) = \lambda\)
Did you know? In a Poisson distribution, the mean and variance are exactly the same! This is a great way to check if a Poisson model is suitable for a set of data.
The Additive Property
If you have two independent Poisson variables, \(X \sim Po(\lambda)\) and \(Y \sim Po(\mu)\), you can just add them together!
\(X + Y \sim Po(\lambda + \mu)\)
Example: If you get 2 emails per hour (\(X\)) and 3 texts per hour (\(Y\)), the total number of notifications per hour is \(Po(2+3) = Po(5)\).
Key Takeaway: Use Poisson when you are "counting" occurrences over a fixed interval.
3. Poisson as an Approximation to the Binomial
Calculating Binomial probabilities with huge numbers is a nightmare. Luckily, if your \(n\) is large and your \(p\) is small, the Poisson distribution can step in to help.
The Rule: If \(X \sim B(n, p)\), you can approximate it using \(Po(\lambda)\) where \(\lambda = np\).
When to use it? Usually when \(n > 50\) and \(np < 5\). It makes your calculations much faster!
4. The Geometric Distribution
The Geometric Distribution, \(X \sim Geo(p)\), is all about waiting for the first success.
Analogy: Imagine you are trying to flip a coin to get a "Head". You keep flipping until you finally get one. \(X\) is the number of flips you needed.
The Probability Formula
\(P(X=x) = p(1-p)^{x-1}\)
Why? Because to get your first success on the \(x\)-th try, you must have failed \(x-1\) times first, and then succeeded once.
Mean and Variance
- \(E(X) = \frac{1}{p}\)
- \(Var(X) = \frac{1-p}{p^2}\)
Memory Trick: If the chance of winning a game is \(1/10\), you'd "expect" to play 10 times to win once. That’s \(E(X) = 1/(1/10) = 10\).
5. The Negative Binomial Distribution
The Negative Binomial Distribution is the big brother of the Geometric distribution. Instead of waiting for the first success, you are waiting for the \(r\)-th success.
Example: A basketball player keeps shooting until they have made 3 baskets (\(r=3\)). \(X\) is the total number of shots taken.
The Probability Formula
\(P(X=x) = \binom{x-1}{r-1} p^r (1-p)^{x-r}\)
Don't panic! Here is how to understand it:
1. The last shot (the \(x\)-th one) must be a success (that's why we have \(p^r\)).
2. In the previous \(x-1\) shots, you must have had \(r-1\) successes in any order (that's the \(\binom{x-1}{r-1}\) part).
3. The rest are failures (\((1-p)^{x-r}\)).
Mean and Variance
- \(E(X) = \frac{r}{p}\)
- \(Var(X) = \frac{r(1-p)}{p^2}\)
Common Mistake to Avoid: In Negative Binomial, \(x\) cannot be smaller than \(r\). You can't get 5 successes in only 3 trials!
Summary Checklist
Before you move on, make sure you can:
- Calculate \(E(X)\) and \(Var(X)\) for any given discrete table.
- Identify Poisson situations (constant rate, independent).
- Use Poisson to approximate Binomial when \(n\) is large and \(p\) is small.
- Distinguish between Geometric (waiting for the 1st success) and Negative Binomial (waiting for the \(r\)-th success).
- Use your calculator to find Poisson and Binomial cumulative probabilities (\(P(X \leq x)\)).
Encouragement: Discrete distributions are all about recognizing patterns. Once you identify which "story" the question is telling (Is it a rate? Is it a wait?), the formulas will fall into place!