Περίληψη:
The method of the nearest neighbors as well as its variants have proven to be very powerful tools in the non-parametric prediction and categorization of experimental measurements. On the other hand, the number of data available today as well as their dimensionality and complexity is growing rapidly in many scientific fields, such as economics, biology, chemistry, medicine, and others. Usually, the data and their characteristics have semantic dependence and a lot of noise. At this point, the sparse data representation that deals with these problems with great success is involved. In this paper we present the application of these two tried and tested techniques for prediction in various fields related to economics. New techniques are presented as well as exhaustive tests for the evaluation of the proposed methods. The results are encouraging to continue research into the possibilities of sparse representation combined with good proven machine learning techniques. © 2021, Springer Nature Switzerland AG.