Assessment of the Level-differentiating Ability of Diagnostic Tools for Predicting the Academic Success of Students

The article substantiates the relevance of developing diagnostic tools for pedagogical forecasting of students' academic success in their field of study. The content, principles of development and requirements for diagnostic tools are considered. The results of testing the tools for predictive profile-differentiated diagnostics of students are described. The following principles were used in developing the tools: competence integrity, prospective continuity, predictability, profile and level differentiating ability, consideration of age characteristics and capabilities of students, diversity, and practical focus. The advantages of the diagnostic tools under consideration are: sufficiently high profile and level differentiating ability, sufficient forecast accuracy with insignificant labor intensity; development in relation to various training profiles according to the identified criteria and indicators of academic success, the possibility of statistical processing and analysis of the results obtained using artificial intelligence technologies. The statistical algorithms used were methods of mathematical statistics and cluster analysis: clustering by the k-means method, hierarchical clustering (merging, tree clustering). The application software package "Statistica" was used to carry out clustering. The data obtained on the basis of cluster analysis methods indicate a sufficiently high level-differentiating ability of this tool and the correctness of its use in the educational process.