УДК 636:004.8:338.5
E.G. Skvortsova
Abstract. The article examines the economic efficiency of synthetic data generation as a tool for overcoming the scarcity of high-quality labeled data in livestock farming. The cost structure of real data collection and preparation is analyzed, including expenses for sensor equipment, expert annotator labor, and losses from unused data potential. Methods for assessing the economic efficiency of synthetic data generation are systematized: comparative cost analysis, data quality metrics, assessment of impact on model productivity, and return on investment (ROI) analysis. Special attention is paid to the economic justification of using generative adversarial networks (GANs), variational autoencoders (VAEs), and procedural generation methods under various data availability scenarios. Based on the analysis of scientific literature and practical cases, key factors determining the economic feasibility of data synthesis are identified: farm scale, expert availability, rarity of target events, and model accuracy requirements. Promising directions for cost optimization are determined, including the development of specialized economic efficiency metrics and the creation of open libraries of synthetic data.
Keywords: synthetic data, economic efficiency, big data, livestock farming, data scarcity, return on investment, precision livestock farming.
Download the full text of the article