Geometric vs Negative Binomial: Key Differences
What you'll learn: You'll understand when to use the geometric distribution versus the negative binomial, and how they're fundamentally related.
The Core Relationship
The geometric and negative binomial distributions are part of the same family—they both model "waiting time" in sequences of Bernoulli trials. The key difference is simply what you're waiting for:
- Geometric Distribution: Counts trials until the first success (r = 1)
- Negative Binomial Distribution: Counts trials until the r-th success (r > 1)
Think of it this way: if you're flipping a coin until you get your first heads, that's geometric. If you're flipping until you get your third heads, that's negative binomial with r = 3.
When to Use Each Model
Use Geometric when:
- You need exactly one success
- Examples: trials until first customer buys, attempts until first free throw made
Use Negative Binomial when:
- You need multiple successes (r ≥ 2)
- Examples: calls until 5 sales closed, games until 10 wins achieved
The Mathematical Connection
The geometric distribution is actually a special case of the negative binomial where r = 1. This means:
- Geometric has parameters: p (success probability)
- Negative binomial has parameters: r (successes needed) and p
Both share the memoryless property when r = 1, but only the geometric distribution is truly memoryless in the classical sense.
Practical Implication
If your problem asks "how many trials until the first success?"—use geometric. If it asks "how many trials until the r-th success?"—use negative binomial. The geometric is simpler to calculate but less flexible.
Key Takeaway: The geometric distribution is the negative binomial distribution with r = 1; use geometric for the first success, negative binomial when you need multiple successes.