The informal description of clustering as finding meaningful groups does not suggest a straightforward way of evaluating clusters or defining an objective function. Many definitions of good clustering exist. Prominently, clustering has been posed as an optimization problem for minimum error (least squared error) or maximum attribute predictability (category utility). However, many proposals have been made to evaluate cluster quality or validity [JMF99]. No consensus has been reached in the community and this dissertation will give a brief overview of evaluation criteria in section 4.4.