Feature selection is an important application of rough set theory in data mining and other fields. How to make incremental feature selection for dynamic information systems is the focus of rough set theory research at present. In incomplete mixed information system, the increasing attribute set is an important form of dynamic change of information system. First, the concept of neighborhood conditional entropy is introduced into incomplete mixed information system, and is represented by matrix method. Then, in view of the dynamic increase of attribute set, an incremental updating method based on matrix form of neighborhood conditional entropy is proposed, and an incremental feature selection algorithm is given based on this incremental updating mechanism. Finally, the experimental results on UCI datasets show that the proposed incremental feature selection algorithm has higher feature selection performance than the non-incremental feature selection algorithm does.