Data Mining in a Nut Shell
In today’s business world, information about the customer is a necessity for a businesses trying to maximize its profits. A new, and important, tool in gaining this knowledge is Data Mining. Data Mining is a set of automated procedures used to find previously unknown patterns and relationships in data. These patterns and relationships, once extracted, can be used to make valid predictions about the behavior of the customer.
Data Mining is generally used for four main tasks: (1) to improve the process of making new customers and retaining customers; (2) to reduce fraud; (3) to identify internal wastefulness and deal with that wastefulness in operations, and (4) to chart unexplored areas of the internet (Cavoukian). The fulfillment of these tasks can be enhanced if appropriate data has been collected and if that data is stored in a data warehouse. According to Stanford University, "A Data Warehouse is a repository of integrated information, available for queries and analysis. Data and information are extracted from heterogeneous sources as they are generated....This makes it much easier and more efficient to run queries over data that originally came from different sources." When data about an organization’s practices is easier to access, it becomes more economical to mine. “Without the pool of validated and scrubbed data that a data warehouse provides, the data mining process requires considerable additional effort to pre-process the data” (SAS Institute).
There are several different types of models and algorithms used to “mine” the data. These include, but are not limited to, neural networks, decision trees, rule induction, boosting, and genetic algorithms.
Neural networks are physical cellular systems which can acquire, store, and
utilize experiential knowledge (Zurada). Neural networks offer a way to efficiently model large and complex problems. Decision trees are diagrams used for making decisions in business or computer programming. Branches are used to represent choices with associated risks, costs, results, or probabilities. Rule induction is a way of deriving a set of rules to classify cases (Two Crows). These set of rules differ from those in a decision tree in that they are independent from one another. Boosting is a technique in which multiple random samples of data are taken and a...
... middle of paper ...
...networks, fuzzy logic and genetic algorithms. http://www.partek.com/
. MIT WINROSA WINROSA is a software tool which generates automatically Fuzzy If-Then Rules from your data. The generated data set can be run by most of the existing fuzzy tools like e.g. DataEngine, fuzzyTECH, and Matlab. http://www.mitgmbh.de/
Attar Software XpertRule Data Mining using high performance parallel SQL technologyA Windows PC client being able to intelligently query the data source on the host server can achieve knowledge Induction. The speed of the process is therefore dependant upon the server - not the speed of the client PC. This allows data mining to exploit the speed offered by MPP servers (Massive Parallel Processors) and database architectures that are optimized for serving queries. http://www.attar.com/
Cavoukian, Ann, Ph.D. “Data Mining: Staking a Claim on Your Privacy.” Jan. 1998
Pryke, Andy. “The Data Mine.” 23 Sep. 1998
SAS Institute Inc. “Data Mining.” 12 Jan. 2000
Two Crows Co. “Introduction to Data Mining and Knowledge Discovery.” 1999
Zurada, J.M. (1992), Introduction To Artificial Neural Systems,
Boston: PWS Publishing Company, p. xv: