A Review on Prediction of Autism using Machine Learning Algorithm

Abstract

      Machine Learning is widely used in many application domains like marketing, education, healthcare, transportation etc. One of the powerful supervised machine learning technique called classification which is commonly used for prediction of class label based on various attributes. Autism is a serious disorder that impacts the development of children and influences communication and interaction ability. So there is a need to develop a system which will accurately classify children with autism in early stages. There are various classification techniques which predict autism with some advantages and disadvantages with respect to performance. This paper surveys various supervised and unsupervised classification algorithms that have used for classification for prediction of Autism Spectrum Disorder (ASD) in patients using various features and proposes a method called ensemble classification for ASD. There are various analytical approaches used for prediction like Bayes network, KNN, logistic regression, decision tree, SVM etc. to improve the accuracy of classifier to higher extent which helps in better decision-making process. The technique discussed in this paper known as ensemble learning helps to improve prediction accuracy of weak classifiers to much higher extent with help of various majority techniques such as voting, averaging etc. which takes decision based on results produced by different classifiers.