Probability theory as a project management tool: bayesian methodology for decision-making

Abstract

This study addresses the problem of inefficient project parameter updating through development of an adaptive decision-making system based on Bayesian theory. Traditional planning methods rely on static estimates that are not systematically adjusted as new data emerges, leading to accumulation of deviations between planned and actual indicators. Analysis of contemporary research demonstrates that Bayesian methods remain predominantly theoretical constructs without sufficient operationalization for project practice. A methodological system has been developed that integrates Bayesian updating into four domains: dynamic risk assessment, adaptive schedule forecasting, quality control optimization, and flexible resource management. The implementation algorithm is structured as a five-stage process from identifying application areas through probabilistic model formation to organizational training. Empirical verification was conducted on a construction facility worth twenty-five million euros, where three Bayesian applications were tested. Meteorological risk assessment reduced the probability of critical delays from twenty-five to seven percent, releasing three hundred thousand euros in reserves. Integration of daily productivity metrics over twenty working days reduced forecast uncertainty from eight to three days, demonstrating a sixty-five percent accuracy improvement. Sequential subcontractor evaluation raised confidence levels from seventy to ninety-five percent, justifying reduction in reserve funds. Synthetic analysis of three updates supported the decision for accelerated facility commissioning with expected value of one hundred nineteen thousand euros at seventy percent success probability. Identified limitations include: dependence on initial assumption accuracy, mathematical barriers for users, operational burden of data collection, misalignment between model assumptions and actual interdependencies. Construction of integrated Bayesian architectures with multilevel parameter linkages has been identified as a prospective direction. Theoretical contribution lies in specification of Bayesian apparatus for project conditions with high dynamics. Applied significance is determined by improved assessment accuracy and substantiation of managerial decisions in infrastructure projects with significant financial risks.