Аналіз ефективності ансамблевих методів та статистичних моделей у системах фінансового планування

Abstract

An approach to building an intelligent personal financial planning system is proposed, which solves the problem of processing heterogeneous data and identifying nonlinear dependencies in users' financial behavior. The application of gradient boosting methods for regression analysis of costs and binary classification of goal achievability is considered. The advantages of using ensemble models over classical statistical algorithms and bagging methods in conditions of unbalanced samples are substantiated.