Data Mining in the Soft Computing Paradigm

Data mining is a set of tools, techniques and methods that can be used to find new, hidden or unexpected patterns from a large volume of data typically stored in a data warehouse. Results obtained from data mining help an organization in more effective individual and group decision-making. Regardless of the specific technique, data mining methods can be classified by the function they perform or by their class of application. Association rule is a type of data mining that correlates one set of items or events with another set of items or events. It employs association or linkage analysis, searching transactions from operational systems for interesting patterns with a high probability of repetition. Classification techniques include mining processes intended to discover rules that define whether an item or event belongs to a particular predefined subset or a class of data. This category of techniques is probably the most broadly applicable to different types of business problems. In some cases, it is difficult to define the parameters of a class of data to be analyzed. When parameters are elusive, clustering methods can be used to create partitions so that all members of each set are similar according to a specified set of metrics. Summarization describes a set of data in compact form. Regression techniques are used to predict a continuous value. The regression can be linear or non-linear with one predictor variable or more than one predictor variables, known as multiple regression. Soft computing, which includes application of fuzzy logic, neural network, rough set and genetic algorithm, is an emerging area in data mining. By studying combinations of variables and how different combinations affect data sets, we develop neural network, a non-linear predictive model that “learns.” Machine learning techniques, such as genetic algorithms and fuzzy logic, can derive meaning from complicated and imprecise data. They can extract patterns from and detect trends within the data that are far too complex to be noticed by either humans or more conventional automated analysis techniques. Because of this ability, neural computing and machine learning technologies demonstrate broad applicability in the world of data mining and, thus, to a wide variety of complex business problems. Rough set is the approximation of an imprecise and uncertain set by pair of precise concepts, called the lower and upper approximations. Each soft computing technique addresses problems in its domain using a distinct methodology. However, they are not substitute of each other. In fact, these soft computing tools work in a cooperative manner, rather than being competitive. This has led to the development of hybridization of soft computing tools for data mining applications (Mitra et al., 2002). It should, however, be kept in mind that soft computing techniques have been traditionally developed to handle small data sets. Extending the soft computing paradigm for processing large volumes of data is itself a challenging task. In the next section, we give a brief background of the various soft computing techniques.