In this paper, models for forecasting the dynamics of demand for products with a short expiration date are constructed. Here we propose to construct models for time series forecasting using the decomposition method and taking into account the assumptions of experts about the influence of certain factors on the behavior of product consumers. These models were implemented and tested on real data, and the obtained forecast were evaluated. The decomposition method and Holt-Winters method were used, and comparative analysis of these models was carried out, and a forecast are made.