企业年报数据中工业增加值异常情况的诊断

长期以来，我国政府对统计数据的质量极为重视，并通过各级统计人员的不懈努力，已经探索出一套较为完整的、准确的和有效的控制数据质量的措施和方法，积累了相当丰富的实际操作经验，对提高统计数据质量起到了必不可少的作用。但是，这些措施和方法更偏重于具体细则的实施和实际经验的运用，缺少控制数据质量方法的介入。特别是在不同经济模式下，统计数据摆动相当显著，仅凭经验判断虚假数据就远远不能满足要求。形势迫切地要求我们除了加大政策和法规实施力度外，还必须尽快引进现代统计方法，实现从经验到经验和方法论相结合的过渡，以…

Diagnosis of Abnormal Phenomena of Industrial Added Value in the Statistical Yearbook of Enterprises

The mixed procedure for identifying multiple outliers is constructed in terms of canonical correlation analysis, cluster analysis and the diagnostic method for multiple outliers in linear model, and then this procedure is applied to the annual report data of enterprises in order to identify whether the concerned industry increase value is abnormal. The advantages of the procedure are easily operational, realizable with software and not intervenient by manual work.