An Entropy-Guided Search Algorithm for Key Factor Combination Analysis in RF/Microwave CAD

Key influencing factor analysis aims to identify influential patterns in interrelated variables. This paper proposes an anytime entropy-guided factor-combination analysis method for discovering feasible multifactor combinations in large discrete or discretized design spaces. The method uses negative factor removal entropy to prioritize dimensions and a best-first search strategy to return valid combinations under a user-specified time budget while preserving multidimensional interactions beyond linear projection methods. It targets engineering datasets in which discrete design choices jointly affect continuous responses, making it suitable for data-driven RF, microwave, and millimeter-wave computer-aided design tasks such as antenna, circuit, and subsystem optimization. Experiments on real-world and synthetic datasets demonstrate that the proposed approach can efficiently produce high-quality factor combinations, achieving fast discovery on moderate-scale data and maintaining practical efficiency on larger-scale settings.