Application Of Improved BP Neural Network In Data Mining

With the rapid development of modern society and the increasingly wide range of Internet applications, the amount of information generated every day is also a sharp increase, especially in network data every day in the increase in the index. In the face of these actual data, how to get useful knowledge from them, more and more people pay attention to and research. Therefore, data mining technology is also to meet the needs of such a huge, and it has been widely used and popularized.Data mining is generally faced with noisy, nonlinear and chaotic data, which is the advantage of the neural network. This paper proposes an improved BP neural network algorithm for data mining research, and through an example to verify the effectiveness of the algorithm in data mining. The research work of this paper mainly has the following parts:1. In view of the drawback that BP neural network structure is lack of theoretical guidance,this paper is proposed to dynamically increase the hidden layer nodes, so that the network can get the appropriate network structure. The improved network model is verified by the two sets of Iris data sets and the national transportation data set. The results show that the improved method can get the appropriate number of hidden layer nodes.2. In view of the drawback that the improved BP algorithm is easy to fall into the local optimum problem, this paper propose an improved BP algorithm based on genetic algorithm.Then, through the sample set of parity problem for verification,the results show that the BP algorithm combined with genetic algorithm can effectively avoid the local optimal solution of the network model.3. The improved BP algorithm is applied in data mining. In the tourism industry data set experiment, the total consumption amount of domestic tourism is predicted and analyzed, the results show that the predicted consumption amount is very close to the actual value. Therefore, it shows that the improved BP algorithm has a very good application value in data mining.