Rough modeling -- a bottom-up approach to model construction (Q2772804)

Data mining methods based on rough set theory extract models which are good for classifying unseen objects and are used for two main purposes: prediction and description. Rough modeling is an approach which in addition is computationally simple enough to handle large data sets. As rough models are flexible in nature and simple to generate, a large number of models can be generated and the best of them can be searched. First experiments indicate that (a) the performance of such rough models is slightly worse than the one of larger rough set induced models but also, that (b) there is strong evidence that rough data models are far more descriptive.