Methods of solving mathematical programming problems and statistical decision-making problems (pattern recognition problems) are presented. Algorithms allowing to take into account the influence of errors of all random variables appearing in the problem (confluent analysis) are considered.