Visualization of Inter-Item Dependency and Importance in Employee Surveys on Well-being

Abstract

Bayesian networks are probabilistic graphical models that represent dependencies between random variables using networks and conditional probability tables. Owing to their ability to concretely analyze inter-variable dependencies, they are utilized in various fields such as healthcare and disease diagnosis, marketing, recommendation systems, and survey analysis. This study organizes data from regular internal employee surveys on well-being using Bayesian networks and refines the resulting model using methods such as multiple regression analysis. Based on the analysis, the study identifies items that significantly impact well-being, revealing that factors such as "vitality" and "trustworthy workplace atmosphere" are of importance. The aim of this research is to organize the dependencies between variables in well-being, visualize the overall structure and items of high importance, and discover effective strategies for enhancing well-being.