The Analysis And Forecasting Of International Exchanging Network Management For Telephone Traffic Based On Data Mining

This thesis utilizes the clustering and classifying technology of data mining to analyze the call data of network management and find points where abnormal calls exist. Mining through association rule to get the inter relationship of caller and receiver in the calling register. We need to forecast the getting-through rate when the international line is busy through BP algorithm and then get the useful information. Detailed contents are as follows:1. Utilizes the K-means algorithm to classify the impropriating times when the international line is busy into five categories. Make classification research of the international outgoing traffic with the ID3 algorithm.Get the decision making tree for situations of abnormal telephone traffic and expatiate main practical examples by using the tree which achieves good results.2. Study the conjunction rule of data mining, make deep analysis of Apriori algorithm and improve the algorithm as well as the preferential rule of international calling in different provinces.3. To get a well-pleasing web model through neural nework and training on the getting-through rate when the international exchange is busy through BP algorithm and using MATLAB. Make use of the web model to forecast the getting-through rate in 5 days, a satisfying result is got. Through analysis and forecasting of international telephone traffic exchange data, some problems of maintenance are solved. Replace the traditional statistics reports for automatically processing will be helpful in guiding maintenance workers to do good job on analyzing web management data of telephone traffic as well as on the improving of automatic level of the machine room.