基于Benford法则与随机数辅助调查法的桂林市统计数据质量评估

发布时间：2018-06-13 05:29

本文选题：统计数据的质量 + 统计数据的“验真”　；参考：《广西师范大学》2017年硕士论文

【摘要】：统计数据是统计工作的基础,没有质量的数据就是一堆毫无用处的数字,而它的存在还是十分有害的,所以统计数据质量的高低决定着统计工作整个过程的成与败。统计数据质量是统计的生命线,能够结合实际情况寻找出一种或者多种提高数据质量问题的合适方法就显得十分重要。统计数据质量的高与低主要体现在数据的准确性、及时性、全面性和系统性四个方面,其中准确性是统计数据质量的核心。准确性是前提,是基础。搞准统计数据是统计人员的天职,也是统计部门的第一任务。本文将从“求真”和“验真”两个方面,采用两种不同的方法对桂林市统计数据的准确性作评估。在数据的“验真”方面即质量检测,本文使用Benford法则对选取的桂林市统计局居民用户生活水平的数据的四个指标进行检测,运用Excel软件对样本数据的首位和第二位数字进行统计检测与分析,结果从直观发现首位数字和第二位数字的分布规律与Benford法则有很好的拟合性,之后分别从相关系数、χ2拟合优度、K-S检测和距离检测几个方面对整体数字进行评估,评估结果表明样本数据的数字分布与Benford法则有很好的拟合性,所以说明被检测的数据是准确可靠的。反过来也说明Benford法则对检测桂林市统计数据质量具有很好的适用性。在数据的“求真”方面即数据的质量控制,介绍了桂林市统计局以大数定律为思想的一种新的数据“求真”方法即随机数辅助调查法,并给出具体的理论证明,然后在其具体的应用方面,选取2013年四大行业随机数辅助调查法产生的F值,使用SPSS软件进行正态性检验,然后进行统计分析,证实了该方法很好的可行性,之后提出了关于F值的调整、具体操作过程的优化以及多种辅助调查方法相结合的建议。最后对论文做出总结和展望,指出本文应用研究的不足,希望以后继续研究补充,以期最大程度的运用这两种方法对桂林市统计数据质量进行评估。
[Abstract]:Statistical data is the basis of statistical work, and data without quality is a pile of useless numbers, but its existence is very harmful, so the quality of statistical data determines the success and failure of the whole process of statistical work. The quality of statistical data is the lifeline of statistics. It is very important to find one or more suitable methods to improve the quality of data according to the actual situation. The high and low quality of statistical data is mainly reflected in the accuracy, timeliness, comprehensiveness and systematicness of the data, among which accuracy is the core of statistical data quality. Accuracy is the premise and the basis. Accurate statistical data is the natural duty of statisticians and the first task of statistical departments. In this paper, two different methods are used to evaluate the accuracy of the statistical data in Guilin from the aspects of "seeking truth" and "verifying truth". In this paper, the Benford rule is used to detect the four indexes of the selected data of the residents' living standard in Guilin Bureau of Statistics. The first and second digits of the sample data are detected and analyzed by Excel software. The results show that the distribution of the first and second digits has a good fit with Benford's rule. Then the whole number was evaluated from the correlation coefficient, 蠂 2 goodness of fit KS detection and distance detection. The results showed that the numerical distribution of the sample data had a good fit with Benford rule. So it shows that the detected data is accurate and reliable. In turn, the Benford rule is applicable to the quality of statistical data in Guilin. In the aspect of "seeking truth" of data, that is, the quality control of data, this paper introduces a new method of "seeking truth" of data based on the law of large numbers in Guilin City Bureau of Statistics, that is, the method of random number auxiliary investigation, and gives the concrete theoretical proof. Then in its concrete application aspect, selects the F value which the 2013 four big industries random number auxiliary investigation method produces, uses the SPSS software to carry on the normality test, then carries on the statistical analysis, confirmed this method is very good feasibility, Then some suggestions are put forward on the adjustment of F value, the optimization of concrete operation process and the combination of various auxiliary investigation methods. Finally, the paper is summarized and prospected, pointing out the deficiency of the application research in this paper, hoping to continue to study and supplement in the future, in order to maximize the use of these two methods to evaluate the quality of statistical data in Guilin.
【学位授予单位】：广西师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：C829.2

【相似文献】