Research On The Method And Application Of Statistical Data Quality Evaluation Based On Multi-source Data

Through learning and drawing on the advanced theoretical results and experience of statistical data quality evaluation methods at home and abroad,the article found that most people conduct statistical data quality evaluation from a single data source,and there are relatively few results and literatures on statistical data quality evaluation from a multi-source perspective.To this end,the article will evaluate and apply the quality of statistical data based on multi-source data.First,comb the theoretical connotation of statistical data quality,and conclude that the statistical data quality connotation of the article is mainly accuracy and consistency;secondly,explain the multi-source data resources to define that the multi-source data of the article comes from different departments Statistical data of the website;then,design the evaluation method for the accuracy and consistency of the data quality:(1)Construct two individual models for the same indicator data from different external sources.(2)On the basis of the individual models passing various tests and the evaluation of the fitting effects,the individual individual models from different sources are optimally combined to obtain their own new combined models.(3)The fitting value is obtained through the optimal model,and the optimal combination model is constructed again to evaluate the accuracy of the statistical data quality.(4)On the basis of the accuracy evaluation,the consistency evaluation of the optimal combination model of the two external sources is then carried out,mainly through the evaluation of the consistency correlation coefficient of the fitted values of the external data from different sources.Finally,the article uses per capita GDP data for applied research.Use the per capita GDP data of the United Nations Statistics Division and the World Bank database from 1978 to 2017 as external data to evaluate the per capita GDP data of the Statistical Yearbook.First,a single model was established for the United Nations Statistics Division: MA(2)model and quadratic exponential smoothing model,and a single model for World Bank data: MA(5)model and quadratic exponential smoothing model was established;Optimal combination model;Then,on the basis of the optimal combination model of per capita GDP data from two external sources,a new optimal combination model was constructed again through fitted values to evaluate the accuracy of the per capita GDP data of the Statistical Yearbook,and it was concluded that 1981,The per capita GDP data for the nine years of 1982,1983,1984,1986,1988,1993,2004,and 2010 are doubtful,and the remaining years are accurate.Finally,the consistency assessment of the optimal combination model of the two sources shows that the data fitting consistency of the GDP per capita from the two sources is very good.