Investigastion Of Method For Predicting House Price Based On BP Neural Network And Baidu Map API

In recent years,with the great increase of people's demand for housing,the rise and fall of housing price has become a livelihood issue that people are increasingly concerned about.Therefore,how to objectively predict the housing price and its change trend has attracted people's attention.Combined with Internet,data mining and other information technology means,it is particularly important to emphatically analyze the factors affecting the housing price.The specific work of this paper includes the following five parts:(1)Preprocessing the original data of real estate data.Preprocessing processes such as missing value processing,outlier processing,and data normalization are performed for each attribute of the original data.(2)Building BP neural network model for house price predicting.In this paper,we build two BP neural network model,respectively,which is based on the original data of BP neural network model and BP neural network based on baidu map API extension data model.The process of model construction is to determine the number of neural network layers,the number of nodes in each layer and the activation function.(3)Baidu map API is used to collect information surrounding the house.The API interface of baidu map is mainly used to expand the original data,and finally obtain macro data such as education,medical treatment and transportation within a certain distance around the house,and use it as the model input,so as to improve the accuracy of the model in predicting the housing price.(4)Model training.In this paper,the neural network toolbox of MATLAB is used to train the above two models,including the setting of model parameters and the graphical analysis of model performance.(5)Model validation,results and application.In this paper,the method of 10-fold cross validation is adopted to train and test the model,and the accuracy of experimental results is more than 70%,which proves that the model construction is reasonable and effective.The two neural network models were compared experimentally,and the experimental results show that compared with the method of price prediction based on original data and BP neural network,the method of predicting price of house price based on both the baidu map API and BP neural network has higher accuracy,and the factors affecting the price of the house are not only the micro factor,but also the macro factor plays a major role.In order to facilitate the use of the model,this paper puts forward the method of "double-layer BP neural network",and adds a BP neural network of rough classification of housing price before the BP neural network of housing price prediction.The research idea of this paper can provide some new methods to study the trend of housing price change,provide a new acquisition method for the macro factors that cannot be quantified,and provide corresponding decision support for home buyers.