The Posterior Distribution
You'll learn how the posterior distribution represents your updated beliefs after combining prior knowledge with observed data.
What Is the Posterior?
The posterior distribution is the end result of Bayesian inference—it's what you actually care about. After observing data, the posterior tells you what you now believe about a parameter's likely values, given both your prior beliefs and the evidence you've collected.
Think of it like updating a weather forecast: you start with yesterday's prediction (your prior), observe today's morning conditions (the data/likelihood), and produce an updated forecast for the afternoon (the posterior). The posterior is your "best guess" distribution that balances what you believed before with what you've just learned.
The Posterior in Bayes' Theorem
Recall Bayes' Theorem:
Posterior = (Prior × Likelihood) / Evidence
Or in probability notation:
P(θ | Data) = P(Data | θ) × P(θ) / P(Data)
The posterior distribution P(θ | Data) is proportional to the prior times the likelihood. The evidence (marginal likelihood) acts as a normalizing constant ensuring the posterior is a valid probability distribution that sums (or integrates) to 1.
Why It Matters
The posterior distribution:
- Incorporates uncertainty: It's a full distribution, not just a point estimate
- Updates belief: Strong data can overcome weak priors; strong priors resist contradictory weak data
- Enables decision-making: You can calculate probabilities, credible intervals, and expected values directly from it
Key Takeaway: The posterior distribution is your updated, data-informed belief about a parameter—the ultimate goal of Bayesian analysis that combines prior knowledge with observed evidence.