COMPARATIVE ANALYSIS OF CLASSIFICATION MODELS FOR DETERMINING THE QUALITY OF WINE BY ITS CHEMICAL COMPOSITION
The relevance of the research is caused by the need to solve problems associated with counterfeit and low-quality products in the wine industry. Despite the comprehensive system of regulation of the turnover of alcoholic products, consumers are still at risk of unconscious purchase of lowquality win...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Tomsk Polytechnic University
2023-03-01
|
| Series: | Известия Томского политехнического университета: Промышленная кибернетика |
| Subjects: | |
| Online Access: | https://indcyb.ru/journal/article/view/14/13 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The relevance of the research is caused by the need to solve problems associated with counterfeit and low-quality products in the wine industry. Despite the comprehensive system of regulation of the turnover of alcoholic products, consumers are still at risk of unconscious purchase of lowquality wine products. Unscrupulous producers prioritize cost reduction over product quality, which negatively affects the overall consumer experience and undermines the reputation of more conscientious wine producers. The latter does not contribute to the development of constructive competition in the market and has a negative impact both on the market as a whole and in terms of ensuring the interests of the consumer. In this context, research activities aimed at an automated objective assessment of the quality of wine in terms of its chemical composition using machine learning methods seem to be relevant. Creating tools that provide a reliable and objective way to distinguish genuine high-quality wines from counterfeit or low-quality counterparts is important to safeguard the interests of consumers and promote constructive competition in the wine market. The purpose of the research is to create a system for automated assessment of wine quality by its chemical composition based on a classification model that provides better compliance with the reference data set. Objects: classification models, including the support vector machine, decision tree, random forest algorithm, neural network, multiple regression and their application for automated wine quality assessment. Methods: machine learning methods for the formation of classification models; statistical methods for assessing the quality of classification and comparing classifiers. Results. Using the reference dataset «Wine_Quality_Data», five alternative solutions were formed based on common multilevel classification models. Using statistical criteria, their complex comparison was carried out. The best solution underlying the automated evaluation system proved to be the solution based on the random forest model. |
|---|---|
| ISSN: | 2949-5407 |