基于Lasso方法和BP神经网络的居民消费支出预测研究

发布时间：2018-07-10 01:04

本文选题：居民消费 + Lasso方法　；参考：《湖南大学》2014年硕士论文

【摘要】：消费是国民经济的重要组成部分，连同投资和出口被形象地称之为“拉动经济的三驾马车”。近些年来，在我国城乡居民收入不断增加，储蓄居高不下和投资过热的同时，居民消费需求却尤显不足，致使我国经济遭受了严重的市场需求约束。在这种情况下，对我国居民未来消费支出进行预测，不仅有助于了解居民消费发展的态势，而且对扩大我国内需和促进国民经济又好又快地发展有一定的借鉴意义。首先，本文在综合考虑影响消费的多种因素的基础上，，运用Lasso方法，分别构建了我国城镇居民和农村居民的消费支出模型，并运用1981-2012年的相关统计数据进行实证分析。实证结果表明：城镇居民可支配收入、通货膨胀水平、利率、少年抚养比等7个变量对城镇居民消费支出影响显著；农村居民可支配收入、通货膨胀水平、社会保障支出等6个变量对农村居民消费支出影响显著。其次，在变量选择的基础上，构建基于Lasso方法和BP神经网络的预测模型，并对我国城乡居民的消费支出进行预测，预测结果显示：基于Lasso方法和BP神经网络的组合预测精度要明显高于BP神经网络、Lasso方法的预测精度，预测结果还显示：在2013-2015年，我国农村居民消费增长率有所提升，城镇居民消费增长率缓慢，城乡居民消费增长率之间的差距呈下降趋势，但是短期内城乡居民消费差距依然难以缓和。
[Abstract]:Consumption is an important part of the national economy, together with investment and export is vividly called "pulling the economy of the troika." In recent years, while the income of urban and rural residents is increasing, saving is high and investment is overheating, the consumption demand of residents is especially insufficient, which makes our economy suffer from serious market demand constraints. In this case, forecasting the future consumption expenditure of our country's residents is not only helpful to understand the situation of residents' consumption development, but also helpful to expand domestic demand and promote the development of national economy. First of all, on the basis of synthetically considering many factors affecting consumption, this paper uses Lasso method to construct the consumption expenditure models of urban and rural residents in China, and makes an empirical analysis by using the relevant statistical data from 1981 to 2012. The empirical results show that seven variables, such as disposable income of urban residents, inflation level, interest rate, juvenile dependency ratio and so on, have a significant impact on the consumption expenditure of urban residents, while the disposable income and inflation level of rural residents have a significant impact on the consumption expenditure of urban residents. Six variables, such as social security expenditure, have a significant impact on rural residents' consumption expenditure. Secondly, on the basis of variable selection, the prediction model based on Lasso method and BP neural network is constructed, and the consumption expenditure of urban and rural residents in China is forecasted. The prediction results show that the prediction accuracy of the combination of Lasso method and BP neural network is obviously higher than that of the BP neural network Lasso method. The forecast results also show that the consumption growth rate of rural residents in China has increased in 2013-2015. The consumption growth rate of urban and rural residents is slow, the gap between urban and rural residents' consumption growth rate is decreasing, but the gap between urban and rural residents' consumption rate is still difficult to ease in the short term.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F126.1

【相似文献】