The split-sample foldover test is an important tool for assessing the accuracy of techniques such as data fusion/integration, ascription and predictive modeling, as it is an empirical procedure that does not depend on unverifiable assumptions. When a split-sample foldover test is conducted, there are many criteria that can be evaluated, some of which are relevant and others not. In this paper, we present a typology of evaluative criteria and we will use a number of applications to illustrate how to choose the relevant criteria.
Symposium: 2003: Cambridge, Massachusetts, Session 6 - Fusion
Authors: De Montigny, Michelle, Soong, Roland
Organisations: KMR Group
Topics: Fusion and Modelling