Response to discussions

Running head: NAÏVE BAYES AND BAYESIAN NETWORKS 2

NAÏVE BAYES AND BAYESIAN NETWORKS 2

Naïve Bayes and Bayesian Networks

Santosh Shrestha

University of Cumberlands

Business Intelligence - ITS-531

Dr. Steve Hallman

July 15, 2020

Naïve Bayes and Bayesian Networks

Naïve Bayes is a machine learning technique that uses a simple probability-based classification method which is derived from Bayes theorem. According to Shadra et al. (2020), the method requires the output variable to have nominal values and although the input variables can be a mix of numeric and nominal types, the numeric output variable needs to be discretized via some type of binning method before it can be used in a Bayes classifier (p. 278). The input variables may have no correlation with each other that contributes to the probability independently. Bayesian networks is a powerful tool for representing dependency structure intuitive way as it reflects the various states of a multivariate model and their probabilistic relationships (Shadra et al., 2020, p. 287). Naïve Bayes is one of the subclasses of Bayesian Networks with the major difference that the Bayesian Network have joined distribution with the occurrences of the events impacting to the prediction.

Process of Developing Bayesian Networks Model

According to Shadra et al. (2020), the Bayesian networks model can be constructed manually with the help of a domain expert or analytically by learning the structure of the network form the historical data by using advanced mathematical methods (p. 288). The popular method for learning the structure of the network is called Tree Augmented Naïve (TAN) Bayes that uses tree structure to appropriate the interactions between predictor variables and the target variable (Shadra et al., 2020, p. 289). The TAN can be constructed by following way: compute the conditional mutual information function for each (i, j ) pair as this function indicates how much information is provided when the class variable is known. Build a complete undirected graph and use a conditional mutual information function to annotate the weight of an edge connecting xi to xj. Build a maximum weighted spanning tree. Convert the undirected graph into a directed one by choosing a root variable and setting the direction of all edges to be outward from it. Construct a TAN model by adding a vertex labeled by C and an arc from C to each xi (Shadra et al., 2020, p. 289). Thus the Bayesian networks model is developed.

Reference

Sharda, R., Delen, D., Turban, E. (2020). Deep learning. Analytics, data science, & artificial intelligence: Systems for decision support (pp. 278-289). NJ, Pearson.