Paper Title
A Rough Set Based Feature Selection Approach For The Prediction Of Learning Disabilities

Abstract
Learning Disability (LD) is a neurological disorder that affects a child’s brain. It causes trouble in learning and using certain skills such as reading, writing, listening and speaking. This paper mainly focuses on the effect of feature selection in the prediction of learning disability in school aged children. The process of feature selection reduces the dimensionality of the data and enables learning algorithms to operate more rapidly and effectively. In this work, as a preprocessing step for classification, significant symptoms are selected from the LD dataset with the help of an RST based feature selection method. The influence of these selected symptoms in the prediction of learning disability is then studied using two popular classifiers MLP and SMO available in Weka Data Mining took kit. Experiments conducted on the LD dataset demonstrate the effectiveness of Rough Set Theory in selecting significant symptoms of LD. The training on the original unreduced data and the results of the comparison established the efficiency of the approach to remove non- significant symptoms from the dataset without affecting the classification performance. Keywords: Rough Set Theory, Feature selection, Discernibility matrix, Reducts, Learning Disability