Data Mining In Mobile Communication Application

With the development of IT industry, the size of the trade data continues to expand. To a reasonable cost to any location to provide high-quality data available. Data Warehouse in the database is put forward on the basis of a new method of data storage and handling, can be described as a form of database. Key data warehouse technology including data extraction, data storage and management, and the three aspects of performance, including data collected data is imported into the warehouse, data on the performance of the major multi-dimensional analysis, mathematical statistics and data mining areas. Multidimensional Analysis of the data warehouse is important manifestations, as MOLAP (Multidimensional OLAP) system is dedicated, therefore, on the field of multi-dimensional analysis tools and products are mostly ROLAP (between OLAP) tool. In practice, the needs of its customers through the statistical data to validate their assumptions on certain things, the final decision-making.Traditional database management system based on the type of data based on the database on the safe and efficient operation and maintenance of large details of data consistency and integrity and the security of the affairs of this operation, on-line analytical processing OLAP is a kind of software technology, it allows analysts, management of information through a variety of possible rapid, consistent and interactive access to the right information to obtain in-depth understanding. OLAP is a data warehouse of display tool, built on the basis of multidimensional views of data, we can offer the user a powerful statistical analysis, statement processing and the ability to forecast trends. Currently, OLAP technology for very active area of research, OLAP deepening understanding.Data mining is with the rapid development of science and technology, the growing size of the database and the people in the database of potential information resources and the demand for rapid development together. Data mining is a large amount of data found in the potential of the technology is application-oriented and more interdisciplinary fields, data mining to promote the wider use of data mining techniques and theoretical research and development, and data mining system is the application of data mining research and bridges, the data mining technology from the promotion to a great role.In-depth study of data warehouse technology, the major database vendors of data warehouse and business intelligence solutions to the comprehensive analysis of the systematic introduction of the decision support system (DSS) the development process, components and the specific realization. OLAP technology to do a more in-depth study and use of third-party OLAP tools, data warehouse on the basis for mobile telecom enterprises constructed decision support systems; Modeling in the early effects not obvious in the manual, the decision to seek the effectiveness and continuity of the system will take the dynamic Decision, the basic strategy is that when the system Mining and reduce the accuracy of the results when the automatic trigger corresponding events, based on customer knowledge warehouse, refresh customer data warehouses, real-learning.Once established model and the historical data verify the reliability of a certain place, you can have access to dynamic scoring process. Users may need to score a new dynamic data, the database can be of a certain part of the score data operation. Including the customers of credit management, fraud prevention strategy management, customer behavior and potential customer analysis and management, trend analysis module structure, which for some modules to handle the data flow analysis, in-depth study of data mining, data mining based on full statistical theory , artificial intelligence, knowledge engineering, neural networks, genetic algorithms and made different levels of learning.Mining new customer service is another key concern of enterprises. Because of its profit is not the main products or services from the initial sales, but from users of a product or service follow-up consumption.Therefore, for these enterprises, how to tap as much as possible to potential customers, attract customers using their services or products is a key step to increase profits. Potential customers found the practice of customers, the customers will be divided into a number of categories. The statistical analysis of customer behavior on the basis of statistical analysis of the various classes of users basic information, identify certain of the basic characteristics of users, then, according to these characteristics corresponding marketing or preferential policies to stimulate the consumption of such customers.Clearly, credit analysis and customer behavior analysis is mining fraud and potential customers, is fundamental to telecom operators enterprise decision support system part of the important function. The work done by this paper is based on the idea of a DSS system.To improve and optimize a data mining system clustering algorithm - multiple clustering algorithm for mining. Traditional systems clustering algorithm for clustering limited to, excessive and clustering indicators indicators rely on the existence of relations between the right circumstances was not high. While multiple systems to improve the clustering algorithm, and clustering samples with variable clustering, first by certain rules indicators divided into several categories, each category contains indicators on the sample data for each cluster, then the second cluster clustering results, in large measure, of the algorithm is indeed degrees.The current high performance, wide application of the principles of classification algorithm C5 done preliminary exploration. The adaptive matched in the training set is generated based on the number of classification models, such as a need for new types of samples, each a score models in the process, and in accordance with different results "voting" is a decision of its kind. The use of cross-validation method, a sample data set group discussions, many Generation Model and verified to be the optimal classification model.