What is a Bayesian Network?
A Bayesian network is a probabilistic graphical model that represents a set of variables and their dependencies/independence through the use of a directed acyclic graph (DAG). Also called Bayes network, belief network, decision network or casual network, a Bayesian network is an interpretable representation of a joint probability distribution.
A Bayesian network uses nodes to represent variables and edges to represent conditional dependencies. Nodes that are not connected represent variables that are conditionally independent.
Figure 1: A simple Bayesian network representing variable 2 as depending on variable 1.
A Bayesian network that models a sequence of variables is called a dynamic Bayesian network. Examples are speech signals and protein sequences. Furthermore, a Bayesian network can be generalized into an influence diagram.
Figure: A simple Bayesian network with conditional probability tables.
Source: AnAj, Public domain, via Wikimedia Commons
Why are Bayesian Networks Important?
Bayesian networks are very important as a tool for understanding dependency among different events, and for determining the probability of one event occurring given that another one already happened, particularly in complex cases.
Bayesian networks are used to build models from data and expert opinions. Their applicability includes event prediction, anomaly detection, diagnosis, reasoning, time series analysis, decision making, and more.
Examples of applications abound. In genetics they are used in generally regulatory network (GRN) models; in computer science, they are applied to document classification, semantic search, spam filters, and error correction; and in medicine, they are an important tool used in diagnosis.
Bayesian Networks + LogicPlum
Bayesian networks are an important tool for prediction and decision making. However, their application requires expert knowledge in areas such as probability theory and graph theory.
LogicPlum’s platform eliminates the need for such expert knowledge by automating the full modeling process, from sample selection to model training and selection. Complementing this, it provides an automatically generated report that describes in detail the steps taken during the model construction.
Guide to Further Reading
For those who want to use R:
Hamed. Bayesian Network in R: Introduction. Available at https://www.r-bloggers.com/2015/02/bayesian-network-in-r-introduction/
For those who prefer Python:
Pomegranate. Bayesian Networks. Available at https://pomegranate.readthedocs.io/en/latest/BayesianNetwork.html