The traditional attribute reduction algorithm based on discriminant matrix can only deal with discrete data, and most of the data contains both discrete and continuous attributes. In response to this problem, this study uses a method that allows discrete data and continuous data to be processed uniformly. This method replaces the original indistinguishable relation with the flexible logic equivalence relation, simplifying the discretization process in the traditional algorithm and improving the efficiency of the algorithm. Experiments show that compared with the traditional algorithm, the improved algorithm omits the process of discretization, and can deal with discrete data and continuous data uniformly.