Study On Effects Of Classified Survey Data With Single Imputation And Class Imputation

The strategy of strengthening the nation by big data, which is hotly discussed during the NPC and CPPCC, is clearly listed in the 13 th Five Year Plan and many representatives suggest to work out the detailed implementation plan to drive the information construction of big data. With the rapid development of big data and the nation transitioning from IT times to DT times, we should hold tightly the policy direction, build big data platform and gradually improve the research technology and method for big data. At present, big data analysis is getting more and more attention of state leaders, experts and specialists, becoming the important project of sectors such as government, culture and enterprise.With the development of practical research, survey data, one of the problems of missing data has become both widespread intractable problem to the statistical analysis experts brought great distress, missing data makes inaccurate results, there is great variation, not a true reflection of the inherent problems, so many experts currently working on about the missing data and the lack of content data interpolation methods, especially multiple imputation techniques.Today, science and education, experts and scholars in the past are no longer confined to a simple approach, but in the continuous improvement, continuous improvement, developing into today’s wide variety of multiple imputation techniques, combined with a variety of data analysis model, expected to achieve better results.The issue surrounding the lack of quality control and survey data values interpolated depth research study will address various aspects, including not only the phenomenon and the reasons for missing data, the survey data itself features related content, including related concepts explained. The core content of this paper is to compare the effect of missing value imputation classification of different types of target variable under different conditions are missing mechanism, which together account for the entire article of four chapters, is the main subject of this research work.The full text includes eleven chapters, the first chapter introduces the basic content, including the background, purpose, literature review and article structure and innovation. The second chapter of the missing data related content, detailing the phenomenon and the reasons for missing data, as well as it used to missing data processing methods. Chapters III and IV describes the specificity and quality control of survey data. Chapter 5 explains the relevant concepts. Chapter VI, VII, VIII, IX study to explore the effect of missing value imputation four different situations. Chapter X is the full text of the conclusions. Chapter XI for references.