All data need to be selected again, the specific option is related to the task. The so-called data warehouse is not an off-the-shelf product that can be purchased. It is a solution for solving problems. The data warehouse uses traditional database technology as the basic means of storage data and management resources, with statistical analysis techniques as an effective way to analyze data and extract information, with artificial intelligence technology as a scientific way to dig knowledge and discovery laws. The establishment of the data warehouse is not to replace the original database, but a new application of database technology for supporting decision analysis. It is because the data warehouse integrates rich massive information, which can greatly simplify the data mining process. According to the China Geological Survey, the Chongqing Port Bureau, Hunan Statistics Bureau, Nanning Local Tax and other data mining is based on data warehouse. Implemented. "Let the data have automatic analysis, judgment, and prediction as a human brain is to establish an analysis model", Wang Shan said: "Modeling is to abstract your professional experience, general rules or universal cases into an analysis model. After you build, you can apply it to those situations, and the result is unknown. "For example, suppose you are a telecommunications company's marketing supervisor, the company wants to develop some new long-distance telephone users. Based on your own experience, when you are looking for who is the most potent new customer, you can first understand which people spend more old customers spending more time on the long-distance call. Because you have a lot of information about old customers, such as age, gender, credit records, and long-distance calls. This is equivalent to you also have the same information on many potential customers. Through statistical analysis of the age, gender, credit records of these old customers, you can infer the most potential new customers. This is much more effective than blindly selling. Modeling is to build a model in the data warehouse, abstract several variables from the specific application. For example, a simplified model of long-distance telephone users can be expressed by customers' occupations, position, annual salary, long calls, gender, and regional variables per month. According to this model, the system can try to dig out the age, gender information of the potential new customer from the large number of call records of the old customers, and help you find new long-distance telephone customers. In fact, the data mining system can be resistant, and ultimately, it is still designed and commanded. The process of excavating data is the process of processing, analyzing, predicting data in accordance with the "model" designed, which is human experience and analysis process is implemented in the computer. Good effect, standard, modeling is the key to talk about the application effect and construction experience of data mining, the relevant person in charge of the China Geological Survey Bureau, Hunan Statistics Bureau, Chongqing Port Bureau and Nanning Taxation Bureau, unanimously believes: the effect is good; if necessary, it is also Future development trend. But the implementation is not easy, the system is still waiting to be improved. Summary experience, they believe that the first data plan must have a unified standard; secondly, modeling is important. According to Zhang Yongbo, the person in charge of the information center data mining project of the China Geological Survey Bureau, from the perspective, first, the analysis efficiency of data mining is much higher than the human evaluation; secondly, data mining can also do the original artificial work, Such as superimposed processing. There are many kinds of geological data. There are dozens of geological data on any space point. Different experts have different evaluations. How to put dozens of data stacks form a comprehensive evaluation, relying on traditional manual operation, not at all It may be realized, and the data mining can be. Therefore, data mining is relatively efficient and complete than manual operation of mineral resources. At the same time, he also believes that it is very difficult to achieve, the hardest is modeling, because it is a process of continuous repeated, continuous improvement. How to make experts experience, thinking, not only need to use professional knowledge, but also use a multidisciplinary theory such as neural network, probability statistics, fuzzy mathematics. In this regard, Xiao Shengli, deputy director of the Data Warehouse Office of Hunan Statistics, also deeply.
