Case studies in the data mining approach to health information analysis
Michael Lloyd-Williams · 1998
It is evident from the studies described that the likelihood of success of a data mining project is to a great extent determined prior to the actual data mining activity taking place, i.e. during the production of the cleaned data. Extreme care should be therefore taken during selection and pre-processing activities in order to ensure that the target data set actually contains relevant and usable data. It is also apparent (and clearly illustrated by the analysis of the infertility databases) that if the data provided does not contain useful information within the context of the focus of the investigation, then the data mining cannot generate such information any more than traditional analysis techniques can. However, it may well be the case that the use of the data mining approach allows this conclusion to be reached more quickly than might ordinarily be the case.