Defining the importance of forecasting in industrial enterprise management using machine learning

In this work the assesement of forecasting importance in making a management decision during an operational control of an enterprise without including the individual cases of dominating impact of forecasted values, including the investing policies and strategic enterprise management was considered. As an alternative to the expert method, a forecast importance assesement method based on the gradient boosted machine learning algorithm analysis of the decision informational field with sub-sequent interpretation of acquired results with Shapley values, allowing for a numerical representation of importance. This method is suggested as a tool to increase efficiency of intellectual decision making supporting systems by the means of including it in the procedure of automated analysis of features' importances. This work is Interdisciplinary and concerns problems of system analysis, economics and psychology.