Artificial Neural Networks as a Means of Restoring Passes in the Initial Data Array

Abstract

The article presents an algorithm for restoration of the original data table using GRNN artificial neural network and the results of the algorithm in testing and empirical data. The given article also deals with calculations of the relative error for different data types with different percentages of passes. The quality of the analyzed data resulting from the passive experiment, as well as the reliability of the analysis results depends on one of the most important factors: the presence of these missing values. The distortion of the original data or incompleteness may distort the result in the general modeling process. Gaps in the original data table may be associated with a complete lack of data (raw data incompleteness) and contradictions arising from the data. And this kind of problem can occur not only with the values of a single attribute, but also with the values of a certain set of attributes especially in those cases when it comes to the large dimension of the factor space.