On the value of data mining tools

Abstract

Improvements in ICTs lead to increasingly high bandwidth becoming widely available, allowing large volumes of data to be moved easily over vast distances. CEOs, CIOs, CFOs and managers in organisations can access increasingly large volumes of data to provide a knowledge basis for making important decisions. As the volume of data grows, making sense it becomes increasingly difficult. Data mining is used to extract useful knowledge from large, fuzzy datasets. There are many different data mining models, such as decision trees, neural networks, clustering, prediction, K-nearest neighbour, and association analysis.Many software vendors have developed data mining tools, based on sophisticated algorithms. To understand how these algorithms work requires considerable technical knowledge that is beyond many IT practitioners. This paper poses the question of how much value such tools are to practitioners who do not have the technical background to fully understand the software and interpret the results.This issue is investigated by comparing two tools based on the decision tree model. Preliminary results suggest that current data mining tools are of limited value to users without considerable knowledge of statistics and data mining.