Strange to say, when building a software cost model, sometimes it's useful to ignore much of the available cost data. One way to do this is to perform data-pruning experiments after data collection and before model building. Experiments involving a set of Unix scripts that employ a variable-subtraction algorithm from the WEKA (Waikato Environment for Knowledge Analysis) data-mining toolkit illustrate this approach's effectiveness.

This article is part of a special issue on predictor modeling.