Application of Artificial Intelligence for Sales Forecasting in Retail

Nowadays, e-commerce became one of the most fast-growing technological markets in Russia. Online shopping is gradually becoming an essential part of the national economy. Over the past decades, the retail industry is dynamically developing and constantly adapting in accordance with the customer needs. So far, however, there has been discussion about the growing trend in online retail and changes in customer behavior. Online shopping is becoming the preferred type of shopping for the most Russians. This trend has been intensified especially during the pandemic, when online stores became the only way to purchase essential products for life. The growing demand has determined the main criteria for leadership in the e-commerce market, such as the presence of a wide region coverage, fast and free delivery by developing logistics infrastructure and accurate demand forecasting. In addition to strict customers’ requirements for companies, there is a high level of competition due to fact that IT companies have entered the market. For the target company in 2021, the priority is to improve operational efficiency. Marketplace plans to focus on the launch and development of products and services so as to increase the supplier base and expand the logistics infrastructure. Moreover, reduction in delivery time is another important theme for online retailers focused on attracting more customers and gaining a competitive edge in the online retail market. This work is aimed at solving such problems of the modern online market as long delivery time to the regions and low accuracy in sales forecasting. The first chapter is devoted to the analysis of the company's activities, its production and financial components. The organizational structure of management and the logistics system of the enterprise are considered. SWOT analysis carried out. In the second chapter, when considering the literary sources, an analysis of the theoretical and practical aspects of the use of artificial neural network models for the problems of forecasting logistics processes is carried out. In the third chapter, solutions are proposed for optimizing the logistics infrastructure (expanding the number of points of orders) based on the recurrent neural network trained for forecasting sales. The bachelor’s thesis is made on 70 pages and contains 29 sources.