DECISION SUPPORT SYSTEM FOR PROJECT RESOURCE PLANNING BASED ON THE RANDOM FOREST METHOD

Abstract

The study develops and justifies the structure of a decision support system (DSS) designed to automate project resource planning processes using the Random Forest method. The relevance of the research is driven by the necessity to transition from subjective estimates to analytical tools for forecasting project costs and duration. The proposed system architecture covers the full data processing cycle: from automated input data collection from corporate databases (such as Jira or MS Project) to the generation of visual reports for management. Implementing the Random Forest algorithm within the DSS framework enables the identification of critical project parameters, specifically technical complexity and external risks, directly at the initiation and planning stages. Special emphasis is placed on the development and implementation of a feature importance visualization mechanism, which transforms the forecasting model into a transparent analytical tool. This allows managers to not only obtain predicted values but also understand the underlying structure of the factors influencing them. It was established that the feature hierarchy, where technical complexity plays a leading role (0.793), enables the project manager to focus on the most critical planning nodes. Such an approach significantly enhances the transparency of decision-making and fosters increased stakeholder trust in the system's recommendations. The practical significance of the results lies in the possibility of implementing predictive management methods. The system identifies potential project bottlenecks before actual difficulties arise, providing the manager with a basis for timely reviews of team composition, budget limit adjustments, or schedule modifications. Thus, the proposed DSS serves as an effective tool for active management, providing decision support to prevent cost overruns and project schedule delays in dynamic environments.