Information System for Monitoring and Forecasting Demand in Retail Food Trade
Students Name: Afanasenko Oleksii Oleksiiovych
Qualification Level: magister
Speciality: Information Systems and Technologies
Institute: Institute of Computer Science and Information Technologies
Mode of Study: full
Academic Year: 2025-2026 н.р.
Language of Defence: ukrainian
Abstract: This qualification thesis presents the concept and implementation of an information system designed for monitoring and forecasting product demand in grocery retail. The primary objective of the project is to develop a tool capable of generating accurate forecasts, providing real-time analytical insights, and supporting managerial decision-making through comprehensive data processing. The relevance of the topic is driven by rapid market fluctuations, increasing competition, growing data volumes, and the rising importance of data-driven operations. In such circumstances, traditional inventory planning approaches become insufficient, which creates the need for systems based on modern forecasting methods and machine learning algorithms. The theoretical part of the thesis examines the nature of food demand, identifies the key factors influencing its fluctuations, and analyses the characteristics of grocery markets in Ukraine and worldwide. Significant attention is given to seasonality, promotional activity, price volatility, and external impacts. Modern forecasting methods are reviewed-ranging from classical statistical approaches to machine learning-based models-which made it possible to justify the methodological choices used in system implementation. Based on the analysis of the domain, a Goal Tree was constructed to systematically decompose the general objective into sub-objectives, tasks, and system requirements. This modelling approach helped define the internal logic of the system, ensure conceptual integrity, and clearly outline expected outcomes. The practical part focuses on designing the system’s architecture and software components. A dedicated module for collecting and processing historical data was developed, incorporating POS-data integration, e-commerce events, warehouse and ERP records, and price information. Special attention was paid to data cleaning, missing value handling, duplicate removal, anomaly detection, and SKU dictionary harmonization. This ensured the formation of a high-quality dataset necessary for reliable forecasting. A core achievement is the development of the demand forecasting module based on a hybrid approach. Seasonal patterns are modelled using SARIMA, which provides transparent and interpretable baseline forecasts. To capture short-term fluctuations and the influence of multiple external factors, the LightGBM algorithm is applied. It processes extensive feature sets including price changes, promotions, weather conditions, local events, and regional differences. As a result, the system produces forecasts that combine statistical stability with the flexibility of machine learning. A user-friendly analytical interface has been designed to present results in an intuitive yet informative way. The implemented dashboards include dynamic sales charts, forecasts, comparisons between baseline and corrected predictions, promo-impact indicators, and scenario-based simulations. This interface allows retail managers to quickly assess the current situation, detect potential shortages or surpluses, understand the causes of deviations, and adjust purchasing or inventory distribution decisions. System testing demonstrated stable algorithm performance, accurate analytical outputs, and a notable reduction in forecasting errors compared to traditional approaches. Additionally, the system enables optimization of logistics costs, decreases losses from overstocking and expired goods, improves service levels, and enhances the overall quality of decision-making. The obtained results confirm the effectiveness of analytical systems in grocery retail and highlight their ability to significantly increase operational efficiency and competitiveness in a dynamic market environment. Keywords – data, data analysis, market, trading References: 1. State Statistics Service of Ukraine. Statistics of consumption and retail turnover. URL: https://ukrstat.gov.ua/ 2. Sachs D., Williamson P. Economic behavior of consumers. - Oxford: Oxford University Press, 2019. - 452 p. 3. Amazon Web Services (AWS). Retail Data Lake Architecture. URL: https://aws.amazon.com/solutions