One of the important datamining function is prediction. Many predictive models can be built for the data. The data may be continous, categorical or combination of both. For either of the above type of data many similar predictive models are available. So it is highly important to choose the possible best accurate predictive model for the user data . For this the models are evaluated using resampling techniques. The evaluated models gives statistical results respectively. These statistical results are analysed and compared . The appropriate model that gives maximum accuracy for the user data is used to do predictions for further data of same type. The predictions thus made by the suitable model can be visualized which forms the decision reports for the user data. A proposal is made to apply fuzzy rough set techniques for evaluation of classifier models [7]. © 2012 Springer-Verlag.