Converting Feature Types in Analysis of Different Types of Data

Abstract

Today there are the large number of methods of data analysis for solving problems of pattern recognition of regression, correlation and factor analysis, which are not applicable in the case of different types of features in the source information. In this paper we propose an approach to solving this problem, named the conversion of feature types. The conversion of feature types is considered as an independent task that allows you to make the transition from non-quantitative features to quantitative ones and in further processing to apply the full range of classical methods of data analysis. The proposed algorithm is implemented in Delphi 10 Seattle the integrated software development sphere. The result of the study was tested when solving the task of recognition of several sets of known data.