Research Of Key Technologies Of Data Mining Tellcommunication Oriented

With the increasingly fierce competition of the telecom industry, Data Mining, as ameasure of the Knowledge Discovery and the Decision Support, has already been widelyused. However, the Human Turbulence and the Inhuman Turbulence on the telecomindustry’s Data Acquisition and Processing System which cause the Data Source’s DataQuality differing largely and some problems of Data Warehouse’s ETL mechanism whichlead to the data missing and data mistake when the data is integrating that result in the lowquality of data which affects the result of Data Mining. So solving the above-mentionedproblems, improving the veracity and the efficiency of Data Mining and making use ofData Mining to improve the normal operations of services in the telecom industry havepractical significance.Although the Data Mining technology has been studied for more than ten years andsome technologies are suggested to try to solve kinds of problems that Data Miningtechnology in the telecom industry is encountered when it is applied, these technologiesdo not adapt to the complex and large-scale data in the telecom industry.This paper take data feature in the telecommunications service as research objects,take improving the efficiency of Data Mining technology in the telecom industry asresearch content, interpret from the standpoint of Data Mining solutions’ practical trait andstudy the Data Mining technology that can meet the various requirements such as theaccuracy, efficiency and Decision Support of Data Mining technology in the telecomindustry in a deep-going way. The innovative work of this paper includes the followingaspects:Firstly, an efficient and accurate data mining program is advanced based on the factthat telecommunications data is unusually large-scaled and the management is extremelycomplex. In order to improve the accuracy of the K-means method of knowledgediscovery, the use of genetic algorithms to optimize the space of the initial value andobtaining the most valuable knowledge in the sub-space by weighting methods areseparately proposed. This algorithm is defined as Valued Kmeans Genetic Algorithm.Secondly, based on the actual needs of the field of telecommunications knowledgediscovery, the Data Mining Grid is an effective means of settlement to the problem that the efficiency of data mining algorithms will continue to fall when the data scale isincreasing in geometric patterns. This paper puts forward a parallel computing schedulingscheme based on Grid Technology and analyzes the scheme’s performance.Lastly, with the rapid development of the3G network, traditional calculationmethods are unable to adapt to the data scene that telecom users’ Network accessbehavior’s data scale increase rapidly dozens of TB. The cloud techniques such as Hadoopplatform are introduced to solve the data storage problem. The appropriate data miningalgorithms are designed from the perspective of practical application. This paper improvesthe traditional decision tree SPRINT algorithms, puts forward a parallel computingprogram and successfully applies to the Hadoop platform.