This study investigates the efficiency of real estate public company using data envelopment analysis. At the same time, the study puts forward the classifying index to classify the effective values solved by the data envelopment analysis method on the condition that the return to the scale is unchanged. Then the study uses the PNN method and MDA method to verify the effectiveness of classification and compares the classification accuracy of the probabilistic neural network (PNN) and the traditional multiple discriminant analysis(MDA) and the result shows that (1)the predicting accuracy of the PNN is better than the traditional MDA; (2)The benchmark enterprise can be concluded through the DEA method; (3) The current listed real estate enterprises overall efficiency value are low and have huge promotion space.