Sample Space Partitioning
In probability theory, a sample space is the set of all possible outcomes of a random experiment. To analyse complex events, mathematicians often break this sample space into smaller, non-overlapping pieces called a partition. A partition of a sample space S is a collection of events B₁, B₂, …, Bₙ that are mutually exclusive (no two share any outcomes) and collectively exhaustive (together they cover every possible outcome). This elegant idea is the backbone of many probability calculations. Partitioning is powerful because it lets us decompose a complicated event A into simpler sub-problems. By knowing how A overlaps with each piece of the partition, we can compute P(A) piece by piece and add the results. This leads directly to the Law of Total Probability, one of the most useful tools in all of statistics and probability. Real-world applications range from medical testing and weather forecasting to machine learning and quality control in manufacturing. Understanding sample space partitioning also builds the foundation for Bayes' Theorem, which allows us to update our beliefs about the probability of a hypothesis when new evidence arrives. Whether you are calculating the chance of a disease given a positive test result, or predicting the reliability of a system with multiple components, partitioning gives you the structured framework to solve the problem systematically.
- Define the Sample Space
- Create Mutually Exclusive Events
- Ensure Collective Exhaustion
- Verify the Probability Sum
- Apply the Law of Total Probability
- Connection to Bayes' Theorem
Define the Sample Space
The sample space S contains every possible outcome of your experiment. For example, rolling a standard die gives S = {1, 2, 3, 4, 5, 6}. Everything we discuss must live inside S.
Create Mutually Exclusive Events
A collection of events B₁, B₂, …, Bₙ is mutually exclusive if no outcome belongs to two different events simultaneously: Bᵢ ∩ Bⱼ = ∅ for all i ≠ j. On a die, define B₁ = {1, 2}, B₂ = {3, 4}, B₃ = {5, 6}. No number appears in two groups.
Ensure Collective Exhaustion
The events must also cover the entire sample space: B₁ ∪ B₂ ∪ … ∪ Bₙ = S. For our die example, {1,2} ∪ {3,4} ∪ {5,6} = {1,2,3,4,5,6} = S. Every outcome is accounted for, so no outcome is missed.
Verify the Probability Sum
Because the partition pieces are mutually exclusive and exhaustive, their probabilities must add to 1: P(B₁) + P(B₂) + … + P(Bₙ) = 1. This is a useful check. For the die: P(B₁) = 2/6, P(B₂) = 2/6, P(B₃) = 2/6, and 2/6 + 2/6 + 2/6 = 6/6 = 1. ✓
Apply the Law of Total Probability
For any event A in S, we can write P(A) = Σᵢ P(A | Bᵢ) · P(Bᵢ). This works because the partition splits A into non-overlapping slices: A = (A ∩ B₁) ∪ (A ∩ B₂) ∪ … ∪ (A ∩ Bₙ), and since these slices are mutually exclusive, their probabilities add up. Each term P(A | Bᵢ) · P(Bᵢ) is the probability that A occurs AND we are in partition piece Bᵢ.
Connection to Bayes' Theorem
Once you have a partition and the Law of Total Probability, Bayes' Theorem follows naturally: P(Bₖ | A) = P(A | Bₖ) · P(Bₖ) / Σᵢ P(A | Bᵢ) · P(Bᵢ). The denominator is exactly the Law of Total Probability applied to P(A). Partitioning transforms a question about a complex conditional probability into a set of simpler multiplications and additions.
The Key Equation
- P(A) — Probability of event A (dimensionless): The total probability of the event A we want to compute, a number between 0 and 1.
- B_i — Partition event i (dimensionless): The i-th piece of the partition of S. All Bᵢ are mutually exclusive and together cover S.
- P(A \mid B_i) — Conditional probability of A given Bᵢ (dimensionless): The probability that A occurs given we know the outcome is in Bᵢ. Defined as P(A ∩ Bᵢ) / P(Bᵢ).
- P(B_i) — Probability of partition piece i (dimensionless): The prior probability of the i-th partition event. These must sum to 1 across all i.
Worked Example
Problem. A factory has three machines producing light bulbs. Machine M₁ produces 50% of all bulbs with a 2% defect rate. Machine M₂ produces 30% with a 5% defect rate. Machine M₃ produces 20% with a 3% defect rate. A bulb is picked at random. What is the probability it is defective?
Answer. The probability that a randomly chosen bulb is defective is 3.1%.
Origins · 1 minThomas Bayes and how this was discovered
Reverend Thomas Bayes was an English statistician and minister who first formulated the theorem that bears his name. His posthumously published essay (1763) introduced the idea of using prior probabilities and updating them with new evidence — a process that rests entirely on partitioning the sample space and applying the Law of Total Probability. His framework became the foundation of Bayesian statistics, now used everywhere from spam filters to scientific research. Bayes never published his famous theorem himself. His friend Richard Price found the manuscript after Bayes' death and sent it to the Royal Society, where it was published two years after Bayes had died.
Related Concepts
- Conditional Probability — The Law of Total Probability uses conditional probabilities P(A | Bᵢ) at its core — understanding conditional probability is essential to using partitions effectively.
- Bayes' Theorem — Bayes' Theorem is the direct extension of sample space partitioning, allowing you to reverse conditional probabilities and update beliefs with evidence.
- Mutually Exclusive Events — Mutual exclusivity is one of the two defining properties of a partition; understanding it is necessary before partitioning can be applied correctly.
- Addition Rule of Probability — The addition rule for mutually exclusive events, P(A ∪ B) = P(A) + P(B), is the algebraic reason why probabilities of partition pieces add to 1.