报告名称：Generalized Beal's Intervals For the Difference Between Independent Binomial Proportions

主办单位：英国立博官网中文版

报告专家：吴延科

专家所在单位：广东海洋大学

报告时间：2020年11月14日15:00-16:00

报告地点：数统学院203室

专家简介：吴延科，广东海洋大学数据科学系讲师，2016年获中国人民大学经济学博士学位。主要从事回归分析、复杂数据建模等方面的研究工作，近年来，在《Computational Statistics》、《Communication in Statistics-Theory and Methods》、《Acta Mathematicae Application Sinica》等期刊发表论文20余篇，公开出版专著和译著4部。

报告摘要：This paper proposes generalized Beal's intervals based on Beal's method for the difference between independent binomial proportions. The new method can efficaciously prevent overt overshoot in theory and in practice. Downward spikes phenomenon can also partly be eliminated through adjusting the weight of the nuisance parameter. The generalized Beal's intervals with the optimal weight have better coverage probabilities and shorter intervals than most of often used ones. Most importantly, the optimal weight value has a simple non-iterative form, therefore there is no much additional computation compared with Beal’s method. Some nice properties are proven and tested by some experiments. A real extreme case is analysed to show the claimed properties and practical usability.

邀请人：刘展