Diomin K.S., Germashev I.V. A Model for Optimizing Feature Space in Regression Problems Based on Medical Data
- Details
- Hits: 26
https://doi.org/10.15688/mpcm.jvolsu.2026.1.3
Kirill S. Diomin
Postgraduate Student, Department of Mathematical Analysis and Function Theory, Volgograd State University
This email address is being protected from spambots. You need JavaScript enabled to view it.
,
https://orcid.org/0009-0002-4571-3437
Prosp. Universitetsky, 100, 400062 Volgograd, Russian Federation
Ilya V. Germashev
Doctor of Sciences (Engineering), Professor, Department of Mathematical Analysis and Function Theory, Volgograd State University
This email address is being protected from spambots. You need JavaScript enabled to view it.
,
https://orcid.org/0000-0001-5507-8508
Prosp. Universitetsky, 100, 400062 Volgograd, Russian Federation
Abstract. The paper presents a mathematical model for optimizing feature space in tasks using machine learning models based on medical data. A hybrid model was proposed that takes into account three main types of feature selection, combining both the assessment of the significance of features and their redundancy, and the quality of machine learning models. It was proposed to use a genetic algorithm to find the optimal subset of features, since a complete search in the tasks under consideration is impossible in an adequate amount of time. An open data set on cancer incidence in the United States of America was used in a computational experiment demonstrating the effectiveness of the proposed model. The initial data set contained 26 features and a target variable, the number of records was 3.047. Different models were trained and tested on the full set of features, the best of which turned out to be a random forest model with a coefficient of determination value of 0.936. As a result of the proposed model, it was possible to optimize the initial set of features to 7, with this coefficient of determination value being 0.917. The resulting model can be used as a tool both for selecting the optimal number of features and for choosing the best machine learning model.
Key words: mathematical models, machine learning, data mining, feature space optimization, regression.
Diomin K.S., Germashev I.V. A Model for Optimizing Feature Space in Regression Problems Based on Medical Data is licensed under a Creative Commons Attribution 4.0 International License.
Citation in English: Mathematical Physics and Computer Simulation. Vol. 29 No. 1 2026, pp. 26-37
 Diomin.pdfURL: https://mp.jvolsu.com/index.php/en/component/attachments/download/1451 | 29 Downloads |