关键词: 多层次推断, 改进估计, 模型推估, 大数据启示

Abstract: Total estimation with multi-level has been the most troublesome problem in promoting sample survey in China's governmental statistics. This paper reviews the existing methods of solving the total estimation with multi-level, and point out the limitations of these methods. Then we propose some solutions from the perspective of improving estimation. This paper explores the methods of using auxiliary information to improve estimation and building statistical model for inference. Meanwhile, we points out the advantages, disadvantages and also the applications of those models. Finally, we discuss some enlightenments of solving total estimation with multi-level in the background of big data.

Key words: total estimation with multi-level, improved estimation, model inference