EM算法在投资组合中的应用  

The EM Algorithm for Portfolio

作　　者：杨卓[1] 吕娜[2] 彭章艳[3] 

机构地区：[1]天津商学院理学院,天津300134 [2]东营职业学院计算机系,山东东营257091 [3]武汉工程大学理学院,武汉430074

出　　处：《天津商学院学报》2007年第3期45-47,共3页Journal of Tianjin University of Commerce

摘　　要：投资者通常要利用历史数据进行投资分析。由于各个证券市场成立的时间不同,投资者所获得的历史数据长度各不相同。如果将较长的历史数据“截断”使其与较短历史数据长度相同,这样将丢失部分有效数据从而影响分析结果,而较长的历史数据将为投资者提供更详细、更有效的信息。本文通过EM算法来解决这个问题。Investors should analyse the histories when they construct the optimizing portfolio. The histories were different because each bond markets appeared in different periods of time Investors would get incompleting data. If the ＂discare＂ information from the time series to make both conger and shorter time period have the same length, they would lose some available data. In fact, the longer histories would tell us more useful information. This dissertation offers EM algorithm.

关 键 词：不完全数据 EM算法 投资组合 

分 类 号：O211[理学—概率论与数理统计]