A Bayesian Network Approach to Study Undergraduates' Brand Consciousness

This study seeks to explore the causal relationship between the factors affecting the level of brand consciousness among local universities students in Malaysia by using Bayesian network. Bnlearn package in R is used for learning the graphical structure of Bayesian networks from a survey data and to perform some useful inferences. The random variables and their conditional dependencies are represented via a directed acyclic graph. The three types of structural learning algorithms used are constraint-based algorithms, score-based algorithms and hybrid structural learning algorithms. The scores from their respective estimated networks will determine the best-fitted network. From our result, score-based algorithm using Hill-Climbing and Tabu Search algorithm provided the best-fitted network for the data. The most significant effect on the level of apparel's brand consciousness among the undergraduates is their state of origin and the year of study.