Improving Advertising Messages by Analysing Marketing Factors, that Determine the Price per Square Meter of Real Estate in Moscow

The paper proposes the stickers’ generation system for real estate listings on Avito.ru classified ads platform. In order to obtain key factors which are used for stickers’ generation the hedonic prices method is used for evaluation of key determinants of real estate price in Moscow, based on data from Cian.ru website using gradient boosting algorithm. In the project I analyze special features of classified ads business model in real estate vertical. It also discusses the key features of marketing communications in the industry. To prepare for the formation of a communication strategy, Kotler’s communication model is used to determine the focal stages of the buyer’s path to the transaction. The study analyzes the key global theoretical studies examining the applicability of the hedonic price method to the residential real estate market. The results of studies from various countries were studied to highlight overlapping factors that remain significant regardless of the specific geography. The global experience has been adapted based on local market research in Moscow and an excess list of potentially significant factors has been formed. As part of the solution of the regression problem, the possibility of applying a variety of algorithms from the basic least squares method to a random forest is considered and the choice of gradient boosting algorithm is justified. Keywords: hedonic price method, gradient boosting, classified, sklearn