The Application Of Prediction Model Based On Linear Regression Method And Neural Network In National Economy Data

“CHINA STATISTICAL YEARBOOK” is a statistical data book which is editedby National Bureau of Statistics of China. The book is mainly responsible for theannual statistics of various data. In this paper, the research object is the second part ofthe book which is the national economy data. The national economy reflectes thenational economic data, including the gross national product, the first industry, thesecond industry and the tertiary industry. Among them, second industry includesindustrial and construction industry. National economic data also includes grossnational income and per capita income. This paper’s aim is studying relationshipsbetween the gross national product and other various numerical, and it studies therelationship through more than20years’ data. The paper built the predictive modelsthrough studying of national economic data.Forecasting is a kind of common method in data mining. In the broad sense, theprediction methods can be divided into linear prediction and nonlinear prediction.Linear prediction methods includes: multiple linear regression method, and nonlinearprediction methods have neural network algorithm. Neural network algorithm is acomputational intelligent algorithm; it mainly depends on the complexity of thesystem. Through the adjustment of neural network system with large numbers ofnodes are connected, it can achieve the purpose of processing information.In this paper, it presents a forecasting model combining linear regression andneural network, and uses the model in studying national economy data. This model’smind comes from the thought combing the linear thought and nonlinear thought, andit is a comprehensive utilization of the thought combing linear and nonlinearprediction.This prediction model realizes the idea by making the linear prediction results asthe input of neural network to of neurons. The Neural network method uses in thepaper is RBF radial basis function neural network. In the prediction model, it uses MATLAB to achieve prediction algorithm.The model represents in the paper includes the following steps:(1) It made multivariate linear regression prediction by sample data informationdata.(2) Using the multiple linear regressions to predicted result as a neuron’s inputlayer of a RBF neural network, and neural networks can get multivariate linearregression prediction results in learning and training process, and it can be bettercombination both multiple linear regressions and neural network.(3) Through RBF neural network, it got prediction results. The results areforecast results combining multiple linear regression and RBF neural network.Comparing neural network prediction algorithm’s results and general multiplelinear regressions’ forecasting results, it proves the correct of the forecast model.