Повышение точности прогноза за счет выделения регулярного тренда и случайного остатка

Abstract

The proposed method is based on calculation of a regular trend deviation. It can be applied to the forecast model accuracy enhancement in regular processes. The method consists in calculation of regular trend after real data sample and forecasting of the trend deviation by combinatorial Group Method of Data Handling algorithm. An example of blood glucose level forecasting in the task of diabetic treatment is shown