In this chapter, we saw how we can represent a complex joint probability distribution using a directed graph and a conditional probability distribution associated with each node, which is collectively known as a Bayesian network. We discussed the various reasoning patterns, namely causal, evidential, and intercausal, in a Bayesian network and how changing the CPD of a variable affects other variables. We also discussed the concept of IMAPS, which helped us understand when a joint probability distribution can be encoded in a graph structure.
In the next chapter, we will see that when the relationship between the variables are not causal, a Bayesian model is not sufficient to model our problems. To work with such problems, we will introduce another type of undirected model, known as a Markov model.