To overcome the disadvantages in the traditional hypernetwork, an optimal class-dependent discretization (OCDD)based fine-grain evolutionary hypernetwork model was proposed, in which a continuous data was discretized as a fine-grain data and was used to evolve a hypernetwork classifier. With OCDD, the proposed fine-grain evolutionary hypernetwork lose less information than the traditional hypernetwork which directly convert a continuous data to a binary data. Moreover, fine-grain discrete data processing within the proposed hypernetwork model improves the quality of the probability estimation of a hypernetwork to the training data sets, and further enhances the performance of the hypernetwork classifier. The experimental results on colon, lung, prostate and Leukemia DNA microarray data sets show that the proposed method has a higher recognition rate and better robustness than the traditional evolutionary hypernetworks.