UDC 004
DOI: 10.36871/2618-9976.2022.07-2.001
Authors
Svetlana V. Prokopchina
Doctor of Technical Sciences, Professor, Financial University under the Government
of the Russian Federation, Moscow, Russia
Abstract
The article discusses the basic principles of the implementation
of DATA SCIENCE technologies in the framework of scientific
and practical aspects. It is noted that the functioning of DATA
SCIENCE technologies is always implemented in conditions
of uncertainty and diversity of data and knowledge.
It is shown that one of the basic principles of DATA SCIENCE
technologies in conditions of uncertainty is the principle of regularization
of data spaces and solutions. To regularize the unified
information space of applied problems, regularization is proposed
by means of multidimensional hierarchical scaling, which
is provided by the use of special types of scales with dynamic
constraints. The implementation of such scaling is based
on Bayesian intelligent technologies.
It is also shown that data science is interdisciplinary, an important
component of the set of disciplines is the metrology
of data and knowledge. The methodology and technologies
of metrological substantiation of data and knowledge are
presented. Complexes of metrological characteristics are
proposed, including indicators of accuracy, reliability, reliability,
risk of decisions, the amount of information received
and/or entropy.
Practical examples of the use of these technologies are given.
Keywords
DATA SCIENCE, Regularization, Scaling, Metrology, Bayesian intelligent technologies
