Overview of data Mining for databases (II.)

Data | How do database data mining tools accurately tell you important information that is hidden in the depths of the database? And how do they make predictions? The answer is modeling. Modeling is actually creating a model when you know the results and applying the model to situations that you don't know about. For example, if you want to look for an old Spanish shipwreck in the sea, perhaps the first thing you can think of is looking for the time and place where you found the treasure in the past. Then, after investigation, you find that most of these shipwrecks are found in the Bermuda Sea area, and that the sea has some characteristics of the ocean currents, and the course of that time also has certain characteristics to find. In these many similar features, you abstract and generalize them into a universal model. With this model, you have a good chance of discovering an unknown treasure in another place with a lot of the same characteristics.

Of course, this method of modeling abstraction has been widely used by people before the advent of data mining techniques and even computers. Modeling in a computer is not much different from a previous modeling approach, and the main difference is that the amount of information the computer can handle is much larger than it used to be. Computers can store a large number of different cases of known results, and then the data mining tools Bisha from these large amounts of information and extract the information that can produce the model. Once the model is established, it can be applied to judgments that are similar but whose results are unknown. For example, now that you're a marketing director for a telecoms company, and you want to develop some new long-distance phone users, are you going to wander through the streets and distribute ads? It's like finding a treasure in the sea without a destination. In fact, it's much more efficient to use your previous business experience to reach out to a customer purposefully than to advertise aimlessly.

As a marketing director, you can know a lot about your customers: Age, sex, credit history, and the use of long distance calls. From a good point of view, mastering the information of these customers is to master the same information of many potential users. The problem is that you don't necessarily know the use of their long-distance calls (because their long-distance calls may be another telecoms company). Now your main focus is on who has more long-distance calls. From the table below, we can abstract some variables from the database and build a model that can be classified and marketed.
Customer potential
General Information
(e.g. demographic data) known
Private information
(e.g. customer transactions) known to be pending

Table Ii. Data Mining applied to classified marketing

Based on the computational model we have created from general information to private information, we can draw information from the table in the lower right table. For example, a simplified model of a telecom company can be: more than 60,000 U.S. dollars of more than 98% of customers, a monthly long-distance more than 80 dollars. Based on this model, we can use this data to infer that the company is not yet clear private information, so that the new customer base can be generally determined. Market-test data for small markets can be extremely useful for such a model. Because the small scope of the test data mining, can be the whole market for classified sales to lay a good foundation. Table III describes another common application of data mining: forecasting.
The past is now the future
Static information and current schedule known known known
Dynamic information known to be pending

Table III. Application of data mining in forecasting

The architecture of data mining

Many of the existing data mining tools are independent of the data warehouse, and they require independent input and output data, as well as relatively independent data analysis. To maximize the potential of data mining tools, they must be tightly integrated with the data warehouse, like many business analytics software. In this way, when people change the parameters and depth of analysis, high integration can greatly simplify the data mining process. The following illustration shows a high-level analysis process in a large database.




The Integrated Data Mining system

Application Data mining technology, the ideal starting point is to start from a data warehouse, this data warehouse should keep all the customer's contract information, and should have the corresponding market competitor's relevant data. Such databases can be databases on a variety of markets: Sybase, Oracle, redbrick, and so on, and can be optimized for speed and flexibility in the data.

OLAP servers on-line analytical systems can enable a very complex end-user business model to be applied to the data warehouse. The multidimensional structure of a database allows users to analyze and observe their business operations from different perspectives, such as product classification, geographic classification, or other critical perspectives. The data mining server must and the online Analysis server in this case, and the data warehouse is tightly integrated, so that it can directly track data and help users make quick business decisions, and users can constantly find better behavior patterns when updating data and apply them to future decisions.

The appearance of the data mining system represents the transformation of the basic structure of the conventional decision support system. Unlike query and reporting languages, which simply feed back data query results to end users, data mining advanced Analysis servers apply the user's business model directly to their data warehouse and give feedback to the user about the results of a related information analysis. This result is an analytic and abstract dynamic view layer that usually varies according to the user's different needs. Based on this view, various reporting tools and visualization tools can present the results of the analysis to users to help them plan what action to take.

A tool for generating profits

Many companies have successfully installed data mining tools. Companies that used the technology earlier are mostly information-intensive companies, such as financial services and mail marketing systems, but the technology is now ready to be applied to companies as long as the company has large databases and a strong desire to improve management through software technology. But using data mining technology, the company must be two key factors, one is a large, integrated database, and the other is a well-defined business process, so that data mining is very tightly applied to the company's data.

With some successful applications of data mining technology, for example, a drug company, by analyzing its recent marketing intensity and sales results, determines which marketing activity has the greatest impact on the high value-added physician population in recent months, based on the competitor's sales activity information and on the local health data system. The drug company can then use its office network, the analysis results are communicated to the local sales representatives, the sales representatives can make the corresponding sales decision according to the key information of the company, so that in the fast changing and dynamic market, the sales representatives can make the best choice according to the analysis of the special situation.

Conclusion
The large Data warehouse, which integrates customers, suppliers and market information comprehensively, leads to the explosive growth of Inerrorformation in the company, which needs to be analyzed in a timely and accurate manner in the market competition. In order to more timely and more accurately make the choice for the enterprise, data mining tools based on relational database and online analysis technology have brought us a new turn. At present, data mining tools are developing at an unprecedented speed, and expanding the user community, in the increasingly fierce market competition in the future, with data mining technology will be faster than others to obtain more rapid response, to win more business opportunities.