Diagnosis of core production dependence on production infrastructure based on ensemble method of machine learning

The rational organization of the production infrastructure of an industrial enterprise has a significant impact on the level of profitability of production. New technologies for integrating business processes are evolving in the face of industrial automation. The aim of the article is to diagnose the level of economic dependence of the core production on the production infrastructure. As a research method, the algorithm of the ensemble method of machine learning “Random Forest” is proposed. The parameters that quantitatively and qualitatively describe the costs of main and auxiliary production, the costs of repair facilities, and the level of production efficiency have been developed. Approbation of the algorithm on the example of chemical enterprises allowed distinguishing three classes of productions by the nature of the organization of production infrastructure and its participation in the core production. The quality of the obtained models is evaluated by calculating the risk of misclassification and the magnitude of cumulative lift, where the class with the most correct classification results is highlighted. Results obtained are primary diagnostics of the organization and capacity utilization of the production infrastructure in order to make decisions on business process restructuring.