statistics

Bayes' Theorem Calculator

Bayes' theorem quantifies how prior knowledge updates when new evidence arrives. Named after 18th-century statistician Thomas Bayes, it's the mathematical foundation for revising beliefs about uncertain events. Engineers use it to diagnose system failures, doctors apply it to interpret medical test results, and data scientists rely on it for machine learning. This calculator computes the posterior probability—the updated likelihood of an event given observed conditions.

Last updated: April 18, 2026

Creators Davide Borchia, PhD

Reviewers Anna Szczepanek and Wojciech Sas

7,022 people find this calculator helpful

Understanding Bayes' Theorem

Bayes' theorem solves a specific probability problem: given that something observable has occurred, what is the probability that an underlying cause or condition is true? This differs from asking the reverse question, which is often easier to measure directly.

The theorem formalizes prior probability (what you believe before new information) and likelihood (how probable the new information would be if your belief were true), then produces a posterior probability (your updated belief after considering the evidence).

Imagine a factory produces widgets on three machines. Machine A produces 50% of widgets and has a 2% defect rate. Machine B produces 30% and has a 3% defect rate. Machine C produces 20% and has a 5% defect rate. If you pull a defective widget from the bin, Bayes' theorem tells you which machine most likely produced it—even though you don't know which machine it came from.

The Bayes' Theorem Formula

The fundamental equation calculates the conditional probability of event A occurring, given that event B has been observed:

P(A|B) = [P(B|A) × P(A)] ÷ P(B)

P(A|B) — Posterior probability—the likelihood of A given that B is true.
P(B|A) — Likelihood—the probability of observing B if A were actually true.
P(A) — Prior probability—your initial belief about the probability of A before observing B.
P(B) — Evidence probability—the total probability of observing B across all possible scenarios.

Multi-Hypothesis Extension for Testing

When hypothesis A can occur in multiple mutually exclusive forms, the denominator expands to account for all pathways that could produce the observed evidence B:

P(B) = P(A) × P(B|A) + P(¬A) × P(B|¬A)

This is invaluable in medical testing. Consider a disease affecting 1% of a population. A test correctly identifies 99% of infected patients but also incorrectly flags 5% of healthy people. If someone tests positive, the actual probability they're infected is dramatically lower than 99%—roughly 17%. The high false positive rate swamps the base rate. This explains why confirmatory testing is crucial in medicine.

Deriving Bayes' Theorem from First Principles

The derivation begins with the definition of conditional probability: the probability of two events both occurring divided by the probability of the conditioning event.

Starting with P(A|B) = P(A ∩ B) ÷ P(B) and P(B|A) = P(A ∩ B) ÷ P(A), we recognise that the intersection probability P(A ∩ B) is the same in both equations.

Rearranging: P(A ∩ B) = P(B|A) × P(A). Substituting this into the first equation yields Bayes' theorem. This derivation shows the theorem isn't an arbitrary rule but a logical consequence of how conditional probabilities relate to joint probabilities.

Common Pitfalls When Using Bayes' Theorem

Misapplying Bayes' theorem leads to flawed reasoning, especially in medical and legal contexts.

Ignoring base rates — The prior probability P(A) often carries more weight than people intuitively expect. A rare disease remains rare even with a positive test. Always anchor to the baseline occurrence rate in your population before weighing evidence.
Confusing the conditional directions — P(B|A) and P(A|B) are not interchangeable. The probability that a person with the disease tests positive differs from the probability they have the disease given a positive test. Swapping these is the 'prosecutor's fallacy' and has wrongly convicted innocent people.
Using incomplete or biased evidence — The formula assumes P(B) is correctly estimated. If your data source is skewed—for instance, only testing symptomatic patients—the evidence probability shifts, invalidating downstream calculations. Ensure your data reflects the real-world context you're modelling.
Forgetting that P(B) must be non-zero — Division by zero is undefined. If P(B) = 0, you cannot have observed B, so the question becomes meaningless. Always verify that your evidence has a non-zero probability before computing the posterior.

Frequently Asked Questions

Why is Bayes' theorem important in machine learning?

Machine learning relies on Bayes' theorem to estimate the probability of a class label given observed features. Naive Bayes classifiers—used in spam filtering, sentiment analysis, and text categorisation—directly apply Bayesian updating. Each new piece of evidence (word in an email, phrase in a review) updates the probability that something belongs to a target category, mimicking how humans learn and refine judgements iteratively.

How does Bayes' theorem differ from the simple conditional probability formula?

The conditional probability formula requires knowing the joint probability P(A ∩ B)—how often both events occur together. Bayes' theorem lets you compute P(A|B) from three simpler quantities: the prior P(A), the likelihood P(B|A), and the total evidence P(B). This is powerful because in many real-world scenarios, measuring the likelihood of evidence under different hypotheses is easier than directly observing joint occurrences. You can reverse cause and effect.

Can Bayes' theorem be applied to more than two events?

Yes. You can partition event A into multiple non-overlapping scenarios and compute the posterior for each. The denominator—total evidence probability—becomes a sum across all partitions, weighted by their priors and likelihoods. This extension is essential for multi-class classification, diagnostic reasoning with several competing diseases, and situations where multiple independent causes could produce the same symptom or observation.

What does 'Bayesian inference' mean, and how does it relate to Bayes' theorem?

Bayes' theorem calculates a single posterior given fixed inputs. Bayesian inference is the iterative process of applying Bayes' theorem repeatedly as new evidence arrives, each time using the previous posterior as the next prior. This recursive updating is how search engines refine results, how recommendation systems learn preferences, and how scientists accumulate confidence in hypotheses as experiments accumulate. It formalises the principle of learning from experience.

Why do medical tests sometimes give misleading results even if they're accurate?

A test can be 95% accurate yet still produce more false alarms than true positives in a rare disease scenario. If the disease affects 1 in 1,000 people and the test is 95% accurate, roughly 50 out of 1,000 healthy people will test false-positive versus 9.5 out of 10 infected people testing true-positive. The prior probability—the rarity of the disease—dominates the posterior. This is why doctors order confirmatory tests and why understanding base rates matters in interpreting any screening result.

How do I decide whether to use Bayes' theorem or the conditional probability definition directly?

Use the conditional probability definition P(A|B) = P(A ∩ B) ÷ P(B) if you already know or can easily measure the joint probability P(A ∩ B). Use Bayes' theorem if you know the prior P(A), the likelihood P(B|A), and the evidence probability P(B), but not the joint probability directly. In practice, likelihoods are often easier to estimate from experimental or observational data than joint occurrences.

More statistics calculators (see all)

5 Number Summary Calculator Midrange Calculator Quartile Calculator Possible Combinations Calculator Exponential Distribution Calculator Percentile Calculator Raw Score Calculator Sampling Distribution of the Sample Proportion Calculator