Prediction On Urban Residents' Consumption Expenditure By Using Lasso Class Methods And BP Neural Network

For a long time,we count consumption,investment and exports as three important premises which make our economy get a fast development.Investment and exports always play a very important role in terms of stimulating the economy development of our country.But compared with the level of investment and exports,our country's consumer spending is always in the low level,and how to promote and guide the residents' consumption better in China has become a growing concern,therefore,our party and government has also repeatedly stressed that we need to expand the level of domestic demand of our country,especially expand the level of consumer demand of our country.So discussed what kinds of factors will impact on China urban residents' consumption expenditure has a far-reaching practical significance.With the development of the big data era,we are very concerned the problem of variable selection,how to screen the information effectively in the large amounts of data is the focus of research now.For study the above problems,this paper make the following work:Chapter 1 expounds the background and research purpose about the Lasso class methods,BP neural network and the urban residents' consumption expenditure,the corresponding research results and the present situation of the domestic and foreign scholars.Chapter 2 respectively introduces the method of Lasso,the Adaptive Lasso(abbreviation A-Lasso),the Elastic Net(abbreviation E-Net)and the Adaptive Elastic Net(abbreviation AE-Net),and introduces the basic theory and related algorithm of these method.Chapter 3 mainly research the nature of the Lasso method,the A-Lasso method,the E-Net method and the AE-Net method in the general linear model,and use numerical simulation method to validate it.Chapter 4 mainly introduces the basic principle of BP neural network,algorithm ideas,algorithm steps,and draw the detailed flow chart.And detailed introduces the BP neural network prediction method,provides the thought of how to use the BP neural network to do forecast later.Finally introduced how to do data preprocessing of BP neural network,provide the theoretical basis of how to determine the structure of the BP neural network.Chapter 5 through using the qualitative analysis to determine some of the variables which influence the urban residents' consumption expenditure,then provides the basis for the empirical analysis later.Then use the Lasso method,the A-Lasso method,the E-Net method and the AE-Net to forecast the urban residents' consumption expenditure respectively.Finally,we respectively use the combination forecast method based on the A-Lasso and BP neural network,based on the AE-Net and the BP neural network to forecast the urban residents' consumption expenditure in China.Chapter 6 summarizes the full text and puts forward corresponding Suggestions.Compared with the merits of combination forecast method to the single prediction methods,and put forward the corresponding suggestion on how to improve the level of urban residents' consumption expenditure of our country.