On an approach to data analysis and visualization in the domain of employee–organization relationships

An increasing number of domains in science and industry rely on the intensive use of data. In such domains, obtaining new knowledge is almost impossible without the use of modern methods of data analysis and visualization. A typical example is the domain of human resource (HR) management. This paper proposes an approach to the application of exploratory data analysis, feature extraction from data, and predictive analytics to determine the relationships between an employee and an organization. Correlation analysis is used to identify relationships between data attributes and assess the strength of these dependencies. Word clouds and conditional feature selection are used during feature extraction. A feature that corresponds to the risk of an employee leaving the organization is implemented. The approach is applied on a nationwide dataset of organization's employee survey and contributes to computer science methods in sociology and HR management.