基于有向图构造的数据流预测算法.pdf

28 5 V ol . 2 8 N o. 5 2011 5 MICROEL ECT R ON ICS COM PU T ER M ay 20 11 由 欣, 印桂生 , 150001) : 数据流的预测技术在金融证券管理网络监控等应用中具有重 的作用. 分析近年来有关数据流预测的技 术, 并做出总结. 提出一种新的数据流预测方法, 将输入的数据流构造成有向图数据结构, 并随着新数据流的到来 对有向图进行维护, 从中提取概率矩阵, 利用马尔可夫模型进行计算, 得出数据流下 一时刻可能的到来值. 最后通 过实验验证其可行性. 结论表明, 该算法的精度和预测成功率都较高. : 数据流预测; 有向图构造; 马尔可夫模型 : T P311 : A : 1000- 7 180 20 11) 05- 0040 - 05 Prediction Algorithm Based on Digraphs Constructing over Data Stream Y OU X in, Y IN Gu ish eng Co lleg e of Computer Science and T echnolog y, Har bin Eng ineer ing U niv ersity , H arbin 150001, China) Abstract: T he for ecasting techniques of data stream pr ov ide such an impor tant ro le in t he applications o f f inancial po rtfo lio management and netw o rk mo nitor ing. T he fo recasting techniques o f data str eam in recent y ear s ar e analy zed and made conclu sion . A new f orecasting metho d of data str eam is pr oposed, w hich constructs t he input data str eam to dig raphs str ucture. W hen new data str eam comes, the dig raphs is maintained. T hen the pro bability matr ix that de scribes the dat a st ream approx imat ely can be receiv ed, and the possible value in the next time of a data str eam can be forecasted by t he use of M arkov mo dels . Finally , the f easibility is ver if ied thr ough ex periment . T he co nclusion indi cates that the ratio of the alg or ithm s pr ecision and success forecasting is high . Key words: data stream predictio n; digr aphs constructing; M arko v models [ 2]