Аннотация:new approach for recommender systems design is discussed. The considered system should rely only on anonymous receipts' data and information about products currently bought by a customer. The preference rating for an arbitrary product is calculated as a classification result of a combined feature description of a product that currently is being bought and products that have been bought previously by the same customer. Two different approaches aimed to calculate such descriptions are proposed. The methods were compared with two other techniques in experiments with real retail data, that is estimating preference rating simply as a product sales rate and using association rules. It was shown by experiments that proposed methods outperform two latter ones in terms of areas under ROC curves.