Abstract
The slag-based compound cementitious material (SBCCM) consists of five components, according to the traditional test methods for trial, the workload is very heavy, the efficiency is low. A least-squares mutual information (LSMI) method was created for feature selection of SBCCM. The paper used the LSMI feature selection to sort the importance of the material components, selected the three components that play an important role in the material, and then used the orthogonal test method to quickly test out the optimal ratio of the material. The LSMI is a regression-oriented method capable of solving difficult probability density problems. It sorts the importance of input features by their dependence on output values, thereby screening out important features. The results showed that the LSMI was an effective feature selection method, and it can greatly reduce the workload in the preparation of the SBCCM.
