Common Discrete Distributions: Bernoulli and Binomial
What you'll learn: How to model yes/no trials and count successes across multiple tries using two fundamental probability distributions.
The Core Idea
Think of flipping a coin: it lands heads or tails. That's a Bernoulli trial — a single experiment with exactly two outcomes (success/failure, 1/0, yes/no). The Bernoulli distribution describes this random variable with one parameter p (the probability of success).
Now imagine flipping that coin n times and counting total heads. That's the binomial distribution — it counts successes across repeated independent Bernoulli trials.
Bernoulli Distribution
- Models a single binary outcome
- Has one parameter: p (probability of success)
- Random variable X takes value 1 with probability p, and 0 with probability (1-p)
- PMF: P(X = 1) = p, P(X = 0) = 1 - p
- Expectation: E[X] = p
- Variance: Var(X) = p(1 - p)
Example: Will it rain tomorrow? X = 1 (rain) with p = 0.3, X = 0 (no rain) with p = 0.7.
Binomial Distribution
- Models the number of successes in n independent Bernoulli trials
- Has two parameters: n (number of trials) and p (success probability)
- Random variable X can be 0, 1, 2, ..., n
- Expectation: E[X] = np
- Variance: Var(X) = np(1 - p)
Example: Flip 10 coins (n=10) with p=0.5. X might equal 6 heads, 4 heads, etc.
Why They Matter in ML
These distributions appear everywhere: classifying spam (yes/no), detecting fraud, modeling click-through rates, and anywhere you need to represent binary outcomes or count successes.
Key Takeaway: The Bernoulli distribution models a single yes/no trial with probability p, while the binomial distribution counts how many successes occur in n independent Bernoulli trials.