Determination of water quality from the standpoint of the trap of non-representativeness: errors that occur when combining data series

Abstract

The errors caused by the non-representativeness of the sample data arrays when they are directly combined are analyzed. Using examples of water quality analysis and the quality of metrological support for production processes, it is shown that even if desirable changes in controlled dependencies are observed in different data groups, changes may still be undesirable when combining these groups. It is shown that the correct unification, preserving the direction of functional dependencies, provides for the introduction of weighting coefficients in such a way that the weighted proportion of the studied ratios in the combined data arrays is the same.