The Calculation Method Of China's Quarterly Unemployment Rate In The Background Of Big Data

In the current situation where overcapacity and the economic situation is increasingly grim,unemployment has started to become a problem of common concern to the community.The importance of unemployment measure is increasing.However,influenced by institutional factors,there are many places is not perfect for our unemployment statistics,it can't completely and accurately reflect our true employment situation.It weakens the monitoring role of unemployment rate in macroeconomic greatly.Based on the background of Big Data,this paper proposed improvement ideas of China's unemployment level measurement from data collection,data analysis and data presentation.Based on Baidu Web search data,using commonly used regression model,simple regression,neural networks,support vector machines,random forests,combining three fold cross-validation technique,discussing the method of estimating quarterly unemployment deeply.It is divided into six parts,specifically organized as follows:The first part is an introduction,mainly including research background and significance,research status,research content and methods,innovation and deficiencies.The second section summarizes the changes and current situation of Chinese unemployment measure.Respectively talking China's unemployment measure development from China's unemployment measurement system changes and the lack of unemployment statistical development.Our unemployment statistics system including:the registered urban unemployment system,urban survey unemployment system and unemployment survey system of Census.There exist the problem of poor timeliness,the statistics object is not comprehensive,lack of depth of the statistical indicators,multiple distortion of statistical data,lack of international comparability of the statistical system and other problems of unemployment measure system.The third part explains how to improve the unemployment rate in large measure data theoretical background ideas.First,briefly introduce the theory of Big Data from the definition and characteristics.Secondly,improvement strategies for traditional unemployment from data processing,data analysis and data presentation.The fourth part introduces the methods and results of estimating and modeling China's unemployment.Selecting the unemployment data and website keyword data firstly,after processing,combining with Chow-Lin interpolation,simple regression,neural networks,support vector regression,random forests and other methods,estimating China's unemployment rate.Based on thinking theoretical calculations and statistical theory,Combining RMSE,MSE index minimum principle,the most effective method for unemployment projections is support vector machines,followed by random forest.The fifth part is the conclusion and recommendations.The following conclusions:First,it is reasonable and feasible for the network data to estimate the unemployment rate.Second,the appropriate approach to design is reasonable to infer that Chinese unemployment rate.Third,the Internet search data is unemployment statistics supplemental data,rather than an alternative.Finally,the department should changethe method of China's unemployment rate from the traditional calculation to the big data estimation.The innovation of this paper is mainly reflected as follows:First,correcting the problem of the large gap between the unemployment data and the actual unemployment data,using the Chow-Lin interpolation to solve the problems such as lowfrequency,small number.Second,optimal results-oriented,it chooses the relatively simple model approach and the best model.Third,the statistical offices should change the method of unemployment rate statistics from "as the basis of statistics " to "basic statistics mainly to large data projections auxiliary "to" big data projections based,network monitoring supplemented by".The shortcomings of the paper is:First,because the special nature of the unemployment,when it happened and when it ends are unpredictable.The article only is the Big Data era government measures should be improved and the unemployment rate improved spreadsheet method.It cannot provide a complete solution.Second,because of the lack of official data as a comparison,the paper calculated results coincide exactly with the actual unemployment rate is unknown.Third,the unemployment rate is only improved annual data from quarterly data,it does not fully play the role of network data.