Application of hierarchical cluster analysis in the training of aviation specialists
Abstract. The article discusses an approach to assessing the quality of the formation of competencies in the relevant disciplines mastered by a student or cadet of an educational institution of civil aviation. The assessment is formed by applying one of the unsupervised machine learning methods – hierarchical cluster analysis. Data collection to assess the cadetsβ performance of a higher educational institution of civil aviation was carried out during the development of the discipline Β«Radio equipment of airfieldsΒ». The experimental study took place during the semester. Tests for the discipline were formed in special Google forms, which made it possible to simplify the process of data collection and provide convenient control over the execution of tests. Data processing and further work with them was carried out in the Jupyter Notebook development environment using the high-level object-oriented programming language Python. In the program for the implementation of clustering, the cluster.hierarchy.linkage method from the SciPy library was used. For a graphical determination of the optimal number of clusters into which the sample should be divided, the Cattell’s scree criterion was used. The described approach makes it possible to single out students into separate groups (clusters) in order to automate the verification of the development of competencies.
Keywords: education, competencies, grade, statistics, machine learning, clustering, cluster analysis, dendrogram, automation, program.
Download article in PDF 857.2 kB