Relevant test set using feature selection algorithm for early detection of dyslexia

Abstract

The objective of feature selection is to find the most relevant features for classification. Thus, the dimensionality of the information will be reduced and may improve classification's accuracy. This paper proposed a minimum set of relevant questions that can be used for early detection of dyslexia. In this research, we investigated and proposed a feature selection algorithm that is correlation based feature selection (CFS) and generate classification models based on five different classifiers namely Bayes Net, Simple Logistic and Decision Table. This paper used dataset collected from a computer based screening test developed consists of 50 questions. The result shows that the new set of question suggested from the feature selection algorithm was significantly achieved 100% accuracy of classification and less time was taken for conducting screening test among students.