研究与开发基于电信大数据的5G网络海量用户复访行为预测模型孙玉娣（江苏经贸职业技术学院数字商务学院，江苏南京211168）摘要：5G网络中的用户会产生大量的访问数据，导致用户复访行为难以精准预测，因此提出基于电信大数据的5G网络海量用户复访行为预测模型。从电信大数据中提取用户上网历史行为特征数据，构建数据集。引入多阶加权马尔可夫链模型，通过计算各阶自相关系数，得到模型权重值，计算模型的统计量。经过分析后得到各阶步长的马尔可夫氏链一步转移概率矩阵，从而实现对5G网络海量用户复访行为的精准预测。实验结果表明，该模型拥有最低的均值误差和标准差，以及最高的精度、查全率、查准率、F1指标，可证明该方法在预测用户复访行为方面有着非常明显的优势。关键词：电信大数据；用户复访行为预测；多阶加权马尔可夫链模型；一步转移概率矩阵；自相关系数中图分类号：TP357文献标志码：Adoi:10.11959/j.issn.1000–0801.2023026Apredictionmodelofmassive5Gnetworkusers’revisitbehaviorbasedontelecombigdataSUNYudiSchoolofDigitalCommerce,JiangsuVocationalInstituteofCommerce,Nanjing211168,ChinaAbstract:Usersin5Gnetworkswillgeneratealargeamountofaccessdata,whichmakesitdifficulttoaccuratelypredictusers’revisitbehavior.Therefore,apredictionmodelofmassive5Gnetworkusers’revisitbehaviorbasedontelecombigdatawasproposed.Theuser’shistoricalonlinebehaviorcharacteristicdatawasextractedfromthetele-combigdatatobuildadataset.MultiorderweightedMarkovchainmodelwasintroduced.Themodelweightvaluewasobtainedbycalculatingtheautocorrelationcoefficientofeachorder,andthestatisticsofthemodelwerecalcu-lated.Afteranalysis,theone-steptransitionprobabilitymatrixofMarkovchainwitheachstepsizewasobtained,soastoaccuratelypredicttherevisitbehaviorofmassiveusersin5Gnetwork.Theexperimentalresultsshowthattheproposedmodelhasthelowestmeanerrorandstandarddeviation,aswellasthehighestaccuracy,recall,precisionandF1indicators,whichcanprovethattheproposedmethodhasaveryobviousadvantageinpredictingusers’revisitbehavior.Keywords:telecombigdata,predictionofusers’revisitbehavior,multiorderweightedMarkovchainmodel,onesteptransitionprobabilitymatrix,autocorrelationcoeffi...
