Knowledge Structure and Data Mining Techniques

Abstract

Considerable research has been done in the recent past that compares the performance of different data mining techniques on various data sets (e.g., Lim, Low, & Shih, 2000). The goal of these studies is to try to determine which data mining technique performs best under what circumstances. Results are often conflicting—for instance, some articles find that neural networks (NN) outperform both traditional statistical techniques and inductive learning techniques, but then the opposite is found with other datasets (Sen & Gibbs, 1994; Sung, Chang, & Lee, 1999: Spangler, May, & Vargas, 1999). Most of these studies use publicly available datasets in their analysis, and because they are not artificially created, it is difficult to control for possible data characteristics in the analysis. Another drawback of these datasets is that they are usually very small.