PREDICTIVE ANALYTICS IN THE TENDER PROCUREMENT SYSTEM: METHODOLOGICAL TOOLS AND LIMITATIONS

The article substantiates the need to move from reactive monitoring of the procurement cycle to proactive management based on predictive models. The authors systematize a wide range of methods — from expert-scenario procedures to hybrid ARIMA-LSTM ensembles and reinforcement algorithms — and demonstrate their applied capabilities for forecasting prices, competition levels, deadlines, and the risk of challenging results. A comprehensive analysis of implementation barriers is presented, including technical data deficiencies, organizational fragmentation of information flows, legal restrictions on processing confidential information, and high capital and operating costs. The directions for development in this area lie in the standardization of machine-readable formats, expansion of cloud computing capacities, integration of registers, and the formation of a single information and analytical space. The findings confirm that the introduction of predictive analytics can improve the economic efficiency of procurement and its transparency, but requires comprehensive institutional support and mechanisms for explainability of models.