Abstract: Weakly labeled incomplete mixed data is a common data type, therefore, attribute reduction for weakly labeled incomplete mixed decision systems is a current research hotspot. This paper improves the definition of attribute importance based on the relative neighborhood discrimination degree, constructs an incremental updating mechanism based on the improved relative neighborhood discrimination degree, and designs an incremental attribute reduction algorithm for attribute set increase in weakly labeled incomplete mixed decision systems. Finally, we select 8 datasets from the UCI database and compare the improved algorithm with other similar attribute reduction algorithms. The experimental results show that the improved algorithm has higher reduction efficiency and classification performance, thus verifying the feasibility of the proposed algorithm.