Current trends in the construction of data models of multidimensional information systems using classification methods

Abstract

This study is devoted to current trends in the construction of models of multidimensional information systems using classification methods. It was determined that in the construction of such models, data preprocessing is an important step with the help of which it is possible to solve the problem of data reduction, as well as to achieve optimal data classification. A database on the academic performance of students of the Technical University of the Bachelor's degree program was selected for the study. The resulting database was used to compare CART, LDA, SVM, KNN and RF classification models using Rstudio software. In this paper, it is noted that one of the current trends in the construction of data models of multidimensional information systems is the use of ensembles of classifiers that combine several models or classification algorithms to achieve more accurate and reliable results. By comparing the classification results obtained, it is possible to determine to what extent the predictions obtained using two different algorithms correlate with each other. In general, modern classification methods strive to improve the accuracy and efficiency of classification, adapt to different types of data and structures, and expand the possibilities of using these models in real applications. The study assumes that, taking into account the development of technologies and the growth of data volumes, further development of classification methods will continue.