Bayesian Networks (BNs)

• Regression models have one dependent variable and multiple independent variables.
• Bayesian networks allow networks of dependencies, both direct and indirect.
• Many interdependent variables are allowed.
• Each variable has a conditional distribution that depends on its parents’ values.
• Conditioning on values of observed variables gives confiditional CDFs for an unobserved variable.