Analysis And Research Of Operation Data Based On Enterprise Mobile Office Automation

At present,MOA(Mobile Office Automation,referred to as MOA)business data generated has reached the GB level,with the deployment of more and more points,the business continues to expand,MOA generated by the increasing volume of data.The emergence of this phenomenon,not only in the MOA,many areas are also facing a huge amount of data,how reliable large data storage and analysis has become the current application and research hot and difficult issues.Although Hadoop can deal with the massive log data generated by MOA business,but it is lacking in data modeling and visualization.This paper uses R language and Hadoop framework can be combined with large data processing,analysis and research of MOA operation data based on Hadoop+Hive+R,try running MOA log mining decision information contained,and will build a visualization system for statistical processing of the results,to improve the MOA service industry has important practical application significance.The specific research and discussion of this paper are as follows:(1)According to Hadoop,Hive and R language their respective strengths,built for the MOA operating data processing Hadoop + Hive + R language cloud environment.(2)According to the characteristics of the MOA log data,the internal structure of MapReduce to deeply understand its parallel computing architecture,designed to meet the input and output of MOA log data read write format,in order to improve the treatment efficiency of MOA.For the MOA log format diversity problem,using Hive for processing,saving the preparation of MapReduce time,improve the development efficiency.(3)In order to be able to exchange information,build a well-known enterprise in domestic communication industry MOA real operation data visualization system based on J2EE using Struts2 MVC,mainly provides the physical resources,customer management,resource management,business development,statistical analysis and statistical functions,R language analysis and prediction results in visual processing.(4)Through modeling of the real operation data of MOA License,according to the R package to establish the second exponential smoothing model,Holt-Winters model and ARIMA model to complete the analysis of time series of License,and then tested the model,obtained remarkable model fitting.Finally in order to verify the validity and reliability of the best model in this paper,by fitting the predicted value of the model was verified by comparing with the actual value,calculate the average relative error and the maximum error value,to choose the best prediction model for License sequences.Through the analysis and research results of MOA operation data,it is shown that the framework designed in this paper can solve the above problems of MO A,and the cloud environment can realize the reliable and reliable storage and processing of MOA log.The visualization system embodies MOA operational data intuitively,Which provides an intuitive data basis for the relevant decision-making personnel.The best forecast model ARIMA model can realize the effective prediction of MOA operational data sequence license.