Development of DSS based on statistical methods for industrial enterprises in conditions of digital production

Digital production allows to reduce the time between an event occurring at an enterprise and the response to it because data collection and analysis with subsequent corrective measures are carried out automatically, without human intervention. The same speed and efficiency of analysis and formulating the response should be maintained for decisions made by personnel, otherwise the flexibility of the entire production system decreases. A solution is using decision support systems, which carry out model calculations and give reasonable recommendations based on relevant information, accelerating the decision-making process and improving its quality. A well-developed information infrastructure for digital production allows to construct decision support systems complementing the existing databases, i.e. DSSs are intended for extracting the data, processing them by the selected procedures and presenting the results in a user-friendly format. This reduces the costs for developing and implementing the system. Calculations can use the data characterizing various business processes of the enterprise, i.e., providing a comprehensive resulting solution. If the calculations performed by DSS are based on mathematical models, the user has to be competent in mathematics to correctly interpret the results obtained; this negatively affects the practical applications of such a system. At the same time, mathematical models can significantly improve the quality of decisions, so this problem should be solved when developing a DSS. We have developed a novel DSS model that detects hidden relationships between different indicators of the enterprise’s activity based on correlation and regression analysis, and, with their help, makes forecasts. The algorithm formulates a set of rules that translate the results of model calculations into recommendations that are understandable to users who are not familiar with the underlying mathematical theory. This expands the scope of practical applications of correlation and regression models for making practical decisions at different levels of the enterprise.