A Data Peprocessing Technique Based on Information Entropy

Neural network is an important research topic in the field of data mining. This paper proposing a data preprocessing method based on information entropy which is directed to the irregular and non-linear data characteristics of hydrological data, and uses an improved model of BP neural network. The model uses the method of calculating the entropy of the range entropy, traversing the input variables, searching and replacing them, making the linear relationship between the variables of the input data set weakened. By this way, the non-linear relationship increasing, and the value of the information entropy increasing, so that it can achieving the purpose of improving the accuracy of hydrological prediction. In the actual prediction, the improved model of neural based on the relevant information entropy is superior to the traditional prediction model.