MACHINE LEARNING ASSESSMENT OF DIGITAL WORKS IN FUTURE TEACHER COMPETENCY MAPPING

Abstract

In the context of educational digitalization, the level of professional training of future teachers is increasingly reflected in their academic work. Presentations, infographics, and visual materials are becoming not only a form of reporting but also indicators of the formation of key competencies among students enrolled in teacher education programs. In this regard, the relevance of their systematic analysis and evaluation is growing. The study involves the use of machine learning methods to analyze the digital academic work of students in teacher education programs. The empirical base consisted of 112 student digital works. The experiment was conducted across four dimensions: digital literacy, visualization, creativity, and pedagogical orientation. Random Forest and AI-based models were used for data processing, allowing simultaneous analysis of structural metadata and visual characteristics of the works. The results demonstrated a high level of correspondence between expert evaluations and machine-generated predictions, particularly in the domains of digital literacy and pedagogical orientation. The assessment of creativity proved to be less precise, confirming the difficulty of algorithmic interpretation of original solutions.

The results do not suggest that machine learning can replace expert judgement. Yet they clearly show that it may function as an additional analytical layer, helping to make the evaluation process more consistent and less dependent on purely subjective interpretation.