| As an important component of national economy,the service industry represents the level of economic development and modernization of a country or region.Statistical data quality plays an indispensable role in statistical work,which can reflect the development and changes of some social economy.The quality of statistics not only has an important impact on its use,but also on government decisions.Based on this,this paper selects the GDP data of a province as the research object.In this paper,three methods of time series model,trend simulation model and linear model are used to construct the data quality evaluation model of a GDP.Outlier detection method based on time series model.First,a second order logarithmic difference sequence was used to stabilize the data.Then,an ARMA model was established for the stationary sequence.The order of the model was determined by AIC and BIC criterion,autocorrelation diagram and partial autocorrelation diagram.An outlier detection method based on trend simulation model.Firstly,the tendency part of the exponential model is determined,and then the residual term is modeled by ARMA model.Finally,the years of the suspected outliers are given by the model.An outlier detection method based on linear model.Firstly,the main factors affecting the GDP of a province are given by correlation coefficient and granger causality test.Then,the final explanatory variables were screened out by stepwise regression method.Finally,a linear model was established to calculate the annual error percentage,and the possible outlier years were preliminarily determined.Based on the above model to calculate the error percentage,build a comprehensive evaluation model of data quality,using the analytic hierarchy process(ahp)and entropy weight method,to determine the weights of index layer,calculated the a province each year comprehensive error percentage of GDP,comprehensive evaluation model are used to determine the outliers year GDP in a province,and analyses the corresponding reasons. |
