Finding patterns in nested samples of objects with dynamically distributed data values

The problem of finding hidden patterns in nested samples of objects, the measurement of feature values was considered. The objects of each sample were divided into representatives of two non-intersecting classes. Using a recursive method, the feature values were divided into intervals within which representatives of one of the classes dominated. The results of the partitioning were used to calculate the stability of features. The presence of restrictions on the set of admissible stability values is explained through the properties of membership functions to fuzzy sets. The results of the computational experiment were used to predict the survival of patients with chronic lymphocytic leukemia, depending on the stages of their examination.

Перевод пока недоступен