Knowledge discovery from low quality meteorological databases

The authors consider a meteorological application for KDD. The formatting of meteorological problems can yield extremely wide databases, abundant with missing values and unreliable data. They show how feature selection can be applied to remove irrelevant fields from the database thus creating a problem of workable proportions for later stages. Simulated annealing is used to extract rules describing the various outcomes and finally the results are analysed in the context of the problem domain.