Knowledge discovery for intelligent structural form optimization based on decision tree with rough set attribute importance gain ratio

Promoting the deep integration of artificial intelligence (AI) technology with engineering structural form optimization through design experience, practical cases, and domain-specific knowledge is critical for innovation in AI-based structural form optimization techniques. Firstly, to address issues of attribute redundancy and optimization challenges in decision trees, the concept of attribute importance gain ratio is proposed, and an attribute reduction method based on rough set attribute importance gain ratio is developed. This method provides support for eliminating redundant attributes and enhancing the efficiency of decision tree optimization. Secondly, considering that the conventional decision tree information gain ratio fails to account for dependencies between decision attributes and conditional attributes, an approach that replaces information entropy with attribute importance gain ratio is established for selecting optimal test attributes. Additionally, a method for branching based on equivalence classes derived from test attributes is introduced. This novel approach provides methodological support for thoroughly capturing intrinsic associations among attributes and improving the precision and accuracy of knowledge acquisition using decision trees. Lastly, the effectiveness of the proposed method is validated in the context of intelligent structural form optimization for high-rise buildings. Practical applications demonstrate that this approach constitutes a new and effective method for intuitively and efficiently acquiring knowledge in intelligent structural form optimization of engineering structures.