No abstract

The brain developmental disorder that affects both behavior and communication is autism spectrum disorder (ASD) considered and recognized as a major medical issue affects the increasing population approximately 0.5%– 0.6%. It is a highly heterogeneous neuro-developmental condition that has severe symptoms with various comorbid disorders. Applied Behavior Analysis involves various methods to understand the change in behavior through therapy. The goal is to identify and increase relevant behaviors which can help to diagnose attributes that affect learning. Data mining as the technology handles such medical grounds to predict by analyzing patterns in huge data sets. The outline of the proposed work is to find the relevant attributes from the dataset by normalizing and ranking the attributes. The CFS subset evaluator using various search methods like best first, greedy stepwise and exhaustive search are used to filter relevant feature from the dataset. The ultimate objective of this paper work is to examine the ASD applied behaviors with subject to normalization and ranking. Applying these to the feature selection methods would help for better understanding on various currently wide spread complex medical condition.

This paper is a study on the various machine learning algorithms in order to perform ASD (Autism spectrum Disorder) as per the DSM-V standards. ASD occurs more frequently among children and in order to diagnose this with better accuracy, the study on binary firefly algorithm, a swarm intelligence based wrapper feature selection algorithm is used to obtain best results with optimum feature subsets. This paper will provide overall result after applying it to all types of machine learning models on supervised learning.