METHODOLOGICAL APPROACHES TO ARTIFICIAL INTELLIGENCE BASED FORECASTING IN SCHOOL MANAGEMENT: AN INTEGRATED FIVE-STAGE CONCEPTUAL FRAMEWORK

Abstract

Forecasting is becoming central to data-informed school leadership, yet the methodological foundations on which an artificial-intelligence (AI)-based forecasting mechanism should be built remain under-theorised. This conceptual article addresses a single guiding question: on what scientific basis should forecasting in school management be constructed? Drawing on scholarship in educational management, general systems theory, developmental psychology, probability theory, and reflective practice, the study identifies five methodological approaches — data-driven, systems, multilevel, probabilistic, and reflexive — and maps each to a distinct stage of the forecasting process. The central argument is that these approaches are not competing alternatives but complementary components of one sequential mechanism: data provide the informational basis; the systems approach reveals interdependencies among indicators; the multilevel approach organises analysis across the student, classroom, and school levels; the probabilistic approach enables prediction under uncertainty; and the reflexive approach converts forecasts into continuously refined management decisions. The principal contribution is a coherent methodological framework that reconceptualises AI-based forecasting as a continuous “prediction–outcome–correction” cycle rather than a one-off statistical estimate. The framework is discussed with reference to the general secondary-education context of Uzbekistan, and it provides the theoretical grounding for a subsequent three-stage forecasting mechanism. Limitations of the conceptual design and directions for empirical validation are outlined.