SOLVING WHEAT SELECTION PROBLEMS USING THE DBSCAN ALGORITHM

Abstract

This article discusses the solution to the problem of identifying complex hidden relationships between varieties by clustering using the DBSCAN algorithm, dividing the features of soft wheat varieties into groups (reflecting the growth and development of the wheat plant, components of fertility and fertility, reflecting the quality of the grain).
Based on the values of the features related to different varieties of
 wheat, in one file created on the basis of experimental files, by dividing them into clusters and sets of noise objects using the DBSCAN algorithm, the problems of identifying wheat varieties that are close to each other in features and varieties that change their features due to abiotic factors, as well as sorting wheat varieties with low grain quality and low fertilities were solved.
The efficiency of clustering by dividing the features of objects into groups is shown using the example of wheat selection. ε-neighboured is determined, reflecting the degree of closeness of objects and the number of objects included in 3 groups of features related to wheat varieties. Clusters and noise objects are found based on the detected values. Based on this, it was possible to draw important conclusions about the wheat varieties that retained (did not retain) their bunch and were identified as noise objects in all 3 groups.