I. Introduction
The approaches of data mining can be differentiated by moreover supervised or unsupervised views of learning. In supervised methods of learning, a superior category of adjustable of pre-specified target is there. Then they necessitate a set of training statistics, which is a chronological, heuristic set example in which the values of the object variable are already given. The process of classification is a familiar task of data-mining. There is a requirement of the examination of the characteristics of a recently represented object in classification. Then it tries to allot it to a single set of predefined classes. The Supervised Learning Methods are useful in solving the problems of classification. MLP, SLA and RBFN are the approaches of representation of supervised learning that are applied to the problems of classification. In order to handle progression of classification jobs, a vital dispute met is to find the best method for a particular type of problem. Such as if we need to catch the relationship between the types of data of input output variables and best performing method. Now, the data mining specialists and that the researchers states that there is not only single universal best-performing technique but also many other methods are there. This means, the various kinds of the methods that have their own merits and drawbacks. Thus, a technique can always accomplish the single best for one single definite problem. But, if there is another problem given, then another method is preferred. This condition is known to be the Selective Superiority. Also, the fact means all the methods of supervised learning that have intrinsic restrictions so as to enhance classification accuracy.