Bayesian networks are directed acyclic graphs that show how random variables depend on one another. In Intro to Probability, they organize conditional probabilities so you can update beliefs with Bayes' theorem when new evidence appears.
A Bayesian network is a probability model built from a directed acyclic graph, or DAG, where each node is a variable and each arrow shows a conditional dependence. In Intro to Probability, the point is not just to draw a pretty diagram, it is to keep track of how uncertainty moves through a system.
Each variable has a probability rule attached to it. If a node has no parents, it is assigned a prior probability. If it has parents, it gets a conditional probability table that says how likely the variable is under different parent outcomes. That makes the whole network a compact way to write a large joint probability distribution without listing every possible case one by one.
The word Bayesian matters because the network is built for updating. Once you observe evidence, you can use Bayes' theorem to revise the probability of other variables in the graph. For example, if a medical test comes back positive, the network can update the probability of disease by combining the disease prevalence with the test's accuracy instead of treating the test result as a final answer.
The directed part tells you the direction of dependence, but it does not always mean cause and effect. Sometimes the arrows reflect causal thinking, like disease leading to symptoms. Other times they just encode a useful probabilistic relationship. What matters in this course is that the graph helps you see which variables matter once you know some evidence and which ones become conditionally independent.
A common move is to read the graph as a shortcut for conditional probability reasoning. If two variables are separated by observed evidence, you may be able to treat them as independent given that evidence. That is why Bayesian networks show up in problems about diagnosis, prediction, and risk, where one piece of information changes the chance of several others at once.
Bayesian networks connect a lot of Intro to Probability ideas in one place: conditional probability, joint distributions, and Bayes' theorem. Instead of calculating every probability from scratch, you can use the graph structure to break a messy situation into smaller pieces.
That matters whenever the problem has several linked variables. A disease test, for example, is not just about the test result. You often need the disease prevalence, the chance of a positive result if someone is sick, and the chance of a false positive if someone is healthy. A Bayesian network keeps those pieces organized so you can update the final probability in a logical order.
This term also helps you read model-based questions more carefully. If you see nodes and arrows, you are not just looking at labels. You are looking at which variables are directly connected, which ones are conditionally independent, and where new evidence should change the answer. That is a big step up from treating probability as isolated formulas.
Bayesian networks also show up in later probability and statistics work, especially when people want to model uncertainty in a system with many parts. Even if your class stays at the introductory level, this idea gives you a clean way to think about how evidence changes belief without guessing or overcounting information.
Keep studying Intro to Probability Unit 12
Visual cheatsheet
view galleryConditional Probability
Bayesian networks are built from conditional probabilities. Each node with parents needs a rule for the probability of that variable given the parent values, so you are constantly using the idea of "probability given evidence" rather than a standalone chance.
Joint Probability Distribution
A Bayesian network is really a compact way to represent a joint probability distribution across many variables. Instead of writing every outcome probability directly, you factor the joint distribution into smaller conditional pieces based on the graph.
Conditional Independence
This is the logic that makes Bayesian networks efficient. If the graph says one variable is independent of another once you know some evidence, you do not have to keep multiplying in extra information that is already accounted for.
Disease Prevalence
Medical examples often start with disease prevalence, which is the prior probability that someone has the disease before any test result. In a Bayesian network, that prior combines with test accuracy and symptoms to update the posterior probability.
A quiz or problem-set question will usually give you a graph, a set of conditional probabilities, and one piece of evidence, then ask you to update the probability of another variable. Your job is to read the arrows correctly, identify the priors and conditional probabilities, and apply Bayes' theorem without mixing up the given evidence and the target event.
You may also be asked to decide whether two variables are conditionally independent once a third variable is known. In those questions, the graph matters as much as the numbers. If the structure cuts off a dependency, you should not treat the variables as directly influencing each other after conditioning.
For word problems, the safest move is to label the evidence first, then work outward through the network. That keeps you from using base rates too late or forgetting the false positive and false negative rates that shape the update.
Bayesian networks are directed acyclic graphs that organize probabilities for connected variables.
Each node uses a prior probability or a conditional probability rule, depending on whether it has parent nodes.
The network lets you update beliefs when new evidence arrives, which is where Bayes' theorem comes in.
The arrows show dependence structure, but they do not always mean direct cause and effect.
In Intro to Probability, the big payoff is that a complex joint distribution becomes easier to reason about.
Bayesian networks are graph-based probability models that show how variables depend on one another through directed edges. In Intro to Probability, they are used to organize conditional probabilities and update beliefs when you observe new evidence.
They use Bayes' theorem to turn new evidence into an updated probability for the variable you care about. The graph tells you which probabilities to combine, and the theorem does the actual updating from prior to posterior.
A probability tree usually tracks one branching sequence of events, while a Bayesian network can represent many connected variables at once. The network is better when the relationships are more complex than a single path and when conditional independence lets you simplify the math.
Start by identifying the evidence, then look at the graph to see which variables are directly connected to it. Use the given conditional probabilities and Bayes' theorem to update the target variable, and check whether any conditional independence lets you ignore unrelated branches.