Probability Distribution: Mapping What's Possible

The mental model

A probability distribution is a function that describes the likelihood of all possible outcomes in an experiment. It's a complete guide to the uncertainty of a random event, ensuring the probabilities of all outcomes sum to 1 (or 100%). It answers the question: 'What could happen, and how likely is each possibility?'

How it works

A distribution is a mathematical function. For a discrete variable with a finite number of outcomes, like a dice roll, it assigns a specific probability to each value (e.g., P(roll=1) = 1/6). For a continuous variable, like height or temperature, it describes the probability of the outcome falling within a certain range using a probability density function. The total probability across all possible outcomes, or the total area under the curve for a continuous distribution, is always exactly 1.

When to use it

Probability distributions are fundamental across engineering. Three common uses are: first, in A/B testing, to model click-through rates and determine if a change is statistically significant. Second, in capacity planning, to model user traffic or API response times to provision resources effectively. Third, in machine learning, where classifiers output a probability distribution over possible labels (e.g., 80% chance this image is a cat, 20% a dog).

When not to use it

The main pitfall is not avoiding distributions, but misapplying them. Do not assume your data follows a specific, convenient distribution (like the normal distribution, or 'bell curve') without verifying it. Forcing your data's model into the wrong shape is a common source of major analytical errors. If a process is deterministic and not random, a probability distribution is unnecessary.

One canonical example

A fair six-sided die. The random phenomenon is the roll. The set of possible outcomes is {1, 2, 3, 4, 5, 6}. The probability distribution is a discrete uniform distribution, where the probability of any single outcome is exactly 1/6. The sum of probabilities for all outcomes is 6 * (1/6) = 1, accounting for all possibilities.