Bayes' Theorem Explained: How to Update Beliefs with Evidence

Added:

Frequentism (Frequentist Inference)

Frequentism is the primary alternative to Bayesian inference in statistics and the philosophy of science. While Bayesianism models probability as a subjective 'degree of belief' that is updated with evidence, Frequentism defines probability strictly as the limit of relative frequencies in infinitely repeated, objective trials. Frequentists argue that assigning probabilities to unique hypotheses is unscientific because it relies heavily on 'prior beliefs'—initial assumptions that can introduce personal bias and lack objective grounding. Instead of updating subjective probabilities, frequentist methods (such as hypothesis testing and confidence intervals) focus on the likelihood of observing data under a null hypothesis. This perspective argues that scientific inquiry should remain strictly objective and observer-independent, challenging the Bayesian view that rational thinking is synonymous with subjective probability updating.

Basic probability concepts, including sample spaces, events, and how to calculate simple probabilities (ratios of desired outcomes to total outcomes).
The concept of conditional probability, specifically understanding the notation and meaning of the probability of event A given event B, denoted as P(A|B).
The difference between independent and dependent events, and how the occurrence of one event can influence the probability of another.
Familiarity with visual aids for probability, such as Venn diagrams, tree diagrams, and contingency tables to represent overlapping events.
Bayesian Inference, specifically how to formulate priors, likelihoods, and posterior distributions to estimate statistical parameters.
Naive Bayes Classifiers, which are fundamental machine learning algorithms used for applications like spam filtering and document classification.
Bayesian Networks (Belief Networks), which model complex probabilistic relationships among multiple variables using graphical structures.
Advanced computational Bayesian statistics, such as Markov Chain Monte Carlo (MCMC) methods, used when mathematical integration becomes intractable.
5.5M views151.4Klikes15:10@3blue1brownOriginal Release: 2019-12-22

Bayes' theorem provides a mathematical framework for updating our beliefs based on new evidence, expressed as P(H|E) = [P(E|H) × P(H)] / P(E), where P(H|E) is the posterior probability of a hypothesis given the evidence, P(E|H) is the likelihood of observing the evidence if the hypothesis is true, P(H) is the prior probability of the hypothesis before seeing the evidence, and P(E) is the total probability of the evidence. The theorem demonstrates that new evidence should not completely determine our beliefs but should instead update our existing beliefs proportionally, taking into account both the strength of the evidence and the relative frequencies of different hypotheses in the population.