Data Expansion by Business-Logic Scaling—an Applied Approach for Data Synthesis
Angel Marchev, Vasil Marchev, Alexander Efremov · 2025
This paper introduces a systematic methodological framework for constructing synthetic datasets to mitigate the limitations posed by insufficient sensitive information in data modeling. The proposed methodology comprises a structured sequence of processes aimed at expanding a limited initial real dataset by integrating relationships extracted from business logic. An algorithm is devised to define associations between variables by applying business constraints and assigning scaling weights. The synthetic dataset undergoes validation through a series of tests that assess the quality and coherence of the generated data, incorporating principles of horizontal synchronization and vertical validation. This ensures the development of synthetic datasets that are both statistically sound and scalable, making them suitable for applications in financial and economic analyses as well as for use in data science.