Research Of Data Mining In Mobile Performance Indicators

Quality of network is the lifeline of telecom operators. It makes a direct impact on thelevel of service and determines consumers' satisfaction. So how to stabilize and improvequality of mobile network is a major issue faced by telecom operators.Data mining is one of most popular research of IT. With data mining, useful informationcan be found in huge amounts of data, information turned into action, action converted intovalue. Then people will achieve social and economic benefits. Telecom operators are typicaldata-intensive enterprises which have accumulated a large amount of valuable information.Therefore, it is important for telecom operators to make full use of historical and current datafrom mobile network. Data mining techniques helps telecom operators to find potentialproblems in network and business operation.Mobile performance indicators measure quality standards of network. It is one of themost critical data from mobile network. In the full-service competition background of China'stelecom industry, telecom operators have to face heavy challenges to make effective controland management of the indicators.This thesis focused research on the integration of data mining and mobilecommunications technology; and deeply studied forecasting and optimization of mobileperformance indicators with data mining.In the forecasting of indicators, this thesis studied time series theory, traditional models, stationary models and non-stationary models. Based on the modeling and applications ofARIMA, the thesis chose ARIMA as the model of time series data mining. Handover-successrate is a typical indicator for example, the thesis defined best model parameters,obtainedpredictive values and took handover-success rate as the verification case. In the field of timeseries data mining, the thesis created a method of the meticulous monitoring. The methodmakes monitoring become more sensitive and effective, and makes hidden trouble easier to befound. In the use of ARIMA, the thesis obtained predictive values with good precision whichprovided the data basis for forecasting and optimization.In the optimization of indicators, this thesis introduced signaling data mining technologyinto mobile performance indicators analysis, providing a new analysis tool for optimization.According to results of signaling data mining, the thesis made a guide of networkmaintenance which optimizes existing resources in order to achieve the transformation ofpassive maintenance to high-quality business performance.This thesis studied data mining in forecasting and optimization of mobile performanceindicators. The result shows the application is advanced, easy-to-use and reliable.