Recently there has been significant development in the use of wavelet network methods in various data mining processes, but how to select the proper structure and parameter of wavelet network is a difficulty. This paper offers a kind of solution, in which the clonal selection Algorithm is proposed and the hierarchical structure is used in the coding scheme, thus a novel algorithm of clonal selection wavelet network is presented, meanwhile the evolution of topologic structure and the parameter learning of the wavelet network can be gotten. To evaluate the performance we apply this algorithm to data mining. After illustrating our method with a representative dataset, simulations show that the data mining based on clonal selection wavelet network achieves the better prediction accuracy than other traditional methods.