因果推断及不完全数据的统计分析

项目名称： 因果推断及不完全数据的统计分析

项目编号： No.10801019

项目类型： 青年科学基金项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 王学丽

作者单位： 北京邮电大学

项目金额： 17万元

中文摘要： 统计因果推断是各种科学研究的一个重要目标。相关和因果的关系讨论历史悠久，尽管我们更关 心因果关系，但统计学家得到的往往都是相关关系而不是因果关系。探讨从海量数据中挖掘因果信息，通过因果网络图探讨数据性能提高的关键因素。本项目研究的目的是研究统计因果推断的理论和方法学，同时与医学、生物信息学等领域交叉开展应用研究。本课题的学术思想是将深入地探讨从相关到因果所必需的各种关键假设和因果作用可识别的充分必要条件，研究多混杂因素的判断准则和多因素交互作用推断，探讨基于纵向研究的因果推断，研究因果网络图的学习问题，探讨缺失数据中识别因果的方法。本课题的理论意义是，用数学模型刻画因果推断所需要的假定，探讨实验研究和观察研究的本质差别，研究的理论结果可用作为生物医学领域、基因网络、计算机通讯网络中性能数据处理提供方法，通过数据的统计分析学习因果网络图，做出合理的因果解释。

中文关键词： 因果推断;混杂因素;因果网络图;不完全数据

英文摘要： Statistics causal inference is an important goal of various science researches.The relationship between cause and correlation has a long history, although we more concern about causal relationship, but the statistician obtains often is the correlational dependence but is not the causal relation. To mine causal information from the mass of data is discussed, and the key factors to improve the performance of data through causal network diagram is discussed. The purpose of this research project is to study the statistics causal inference theory and methodology, and communication science, bioinformatics, and other fields in cross-applied research. The academic thinking is to thoroughly discuss the necessary key assumptions from correlation to causal and the necessary and sufficient condition about identifiable causal effects, research on multiconfounders and the interaction of multiple factors , study the Causal inference based on a longitudinal study , study the causal networs and the identification of causal effects with missing data. The theoretical significance of this project is to depict the required assumptions inferring Causal by the mathematical model, and to discuss essential differences between experimental study and observation study. The theoretical study results can be used to improve performance of data-processing methods in the biomedical field, gene networks, mobile communications networks, and to make reasonable causal explanation through the data collected and analyzed.

英文关键词： Causal inference；Confounder；Causal networks; Incomplete data