UDC004.021
INTERPRETABLE MULTICLASS CLASSIFICATION METHOD
A. A. Bubnov, Ph.D (in physics and mathematics.), associate professor, RSREU, Ryazan, Russia;
orcid.org/0000-0002-8022-456X, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
M. A. Gusev, master student, BMSTU, Moscow, Russia;
orcid.org/0000-0002-6067-8994, email: This email address is being protected from spambots. You need JavaScript enabled to view it.
K. A. Maikov, Dr. in technical sciences, full professor, BMSTU, Moscow, Russia;
orcid.org/0000-0003-1864-2397, email: This email address is being protected from spambots. You need JavaScript enabled to view it.
A. N. Pylkin, Dr. in technical sciences, full professor, RSREU, Ryazan, Russia;
orcid.org/0000-0001-9925-2870, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The problem of multiclass classification is studied. The aim is to develop an interpretable method which solves the given problem. Subject area analysis was conducted, resulting in selection of prototype method. It’s binary classification method CORELS. Its limitation, such as inability to carry out a multiclass classification is explored and means to overcome it are offered. As a result, an interpretable method which solves the multiclass classification problem and overcomes prototype limitations is developed. It uses sample as input. Every sample element as in the prototype consists of the list of binary feature values and class label. A study was conducted, as a result of which the results of the classification were found to coincide with the prototype method in the case of a two-class classification and the performance of the modified method in the case of a multi-class classification was checked.
Key words: multiclass classification, interpretable method, branch and bound method, CORELS, binary classification, decision trees, binary classification, bit strings.