A Swarm Based Symmetrical Uncertainty Feature Selection Method for Autism Spectrum Disorders

Autism Spectrum Disorder (ASD) is emerging as a difficult neurological disorder that have a lifetime impact on the development of different skills and talents. Recently, ASD is widely spread among many adults and children because of their food habits, changes in an environment, etc. Data Mining methods are effectively used to identify the perfect features of ASD among children and adults. Feature selection (FS) techniques are necessary for dealing with different dimensional datasets that may incorporate features in the high, little and, medium dimensions. In this paper, a comparative study of several filter feature selection techniques is utilized to diminish the size of the ASD Children dataset. Feature selection methods like SU, IG, CS and optimization technique like PSO, GA and ACO have utilized and proposed a swarm based Symmetrical Uncertainty feature selection (SSU-FS) method based on SU and PSO. For evaluating the Swarm based Symmetrical Uncertainty feature selection method (SSU-FS), classification techniques like Naïve Bayes and ANN have used.