An Algorithm for Constructing Dynamically Stable Classification Trees under the Conditions of Unsteady Financial and Economic Parameters of Objects
Aleksey Myshlyanov, Igor Temkin · Vestnik MEI · 2025
This article describes an original algorithm for constructing decision trees for the classification problem, which takes into account the stability of the target variable and the unsteady nature of the data in the leaves. The existing solutions are reviewed, and the shortcomings of the known approaches are discussed within the review framework. The basic principles of the algorithm operation, its limitations and advantages, and the features of working with data are described.