Testing Attribute Selection Algorithms for Classification Performance on Real Data

We have developed an e-learning platform which records all activity done by students. We try to create an analysis framework that may be used for proving whether or not the platform can classify the students according to their accumulated knowledge. In this process, attribute selection is one of the major initial steps that must be accomplished. In this paper we show what implications may have attribute selection for classification performance and which are the best techniques that may lead to concluding results