CLUSTERING ENGINEERING VACANCIES: K-MEANS AND GAUSSIAN MIXTURE MODEL

Abstract

This article presents an analysis of vacancies for engineers posted on the "Job of Russia" portal. The purpose of the study was to determine the typology of vacancies based on the analysis of the maximum and minimum offered salaries using clustering methods. K-means and the Gaussian mixture model (GMM) were selected as clustering methods. The optimal number of clusters was determined using the "bent knee" method. The results of the study showed that the Gaussian mixture model is a more reliable clustering method compared to the K-means method, since it allows for more accurate identification of clusters with different salary characteristics. The results obtained can be used for further analysis of the labor market and development of recommendations for improving the quality of data and hiring processes.