A Decision Model for Data Mining Techniques

Abstract

Data mining is the process of extracting useful information from very large data sources. Data mining techniques have proven to be very useful in many domains. However, there is no single algorithm or technique that works best across all types of datasets and problems, and it remains "an art" to decide what data mining technique to use for a specific situation. This paper surveys several data mining techniques that can be applied to different business problems, and presents a decision model in the form of a series of 15-20 questions that help identify the best approach or approaches to a specific problem at hand. For some sets of answers, a small number of techniques are dominant. The decision model is based on a review of the current literature, as well as expert experience. The fraud detection problem is adopted as a case study and applied the data mining techniques to draw the insights. We also discuss the applicability of specific techniques to common business in finance, marketing, and business operations.