Abstract
The problems of reliable operation of ERP systems are described. The influence of the human factor on failures and malfunctions in information processing is studied. It is proposed to move from the methodology of statistical analysis of human errors to the methodology of early warning of problem situations. To solve the problem of increasing the reliability of ERP systems, the methods of data mining are used, in particular the method of constructing associative rules. The principles of constructing a special software module aimed at preventing errors are developed. An analysis of the effectiveness of the method is carried out, which confirms the feasibility of its widespread use in the design and operation of ERP systems.