Ways to identify errors in large arrays of numerical information

Abstract

In this article are suggested methods for detecting errors in the numerical information. It is pointed out that in large arrays of numerical data it is impossible to visually highlight data sources with false information and to determine accuracy of which indicators is questionable and requires validation. The author shows that the «compression» of the original information by calculating the correlation matrices, average values of the correlation coefficients, coefficients of skewness and variation facilitates error detection in large arrays of numeric data. If there is a statistically significant regression equation quality check of the source and new array of numerical information can be carried out by comparing the predicted and actual values of the response in a regular array of numerical data with subsequent validation of observations with maximum values of residues.