Autism Spectrum Disorder (ASD) is a neurodevelopmental disability that exhibits sluggish progress in vocal development, restricted interest in normal activity and repetitive disoriented behavior. This syndrome, has gained a lot of attention due to its prevalence among children across all countries and from different economic backgrounds. However, ASD detection and treatment yet remains in its infancy due to the lack of awareness among parents, limited screening of proper developmental milestones and a dearth of diagnostic tools to classify this syndrome with convincing accuracy. Recent studies report that scalable biomarkers for early detection have made little progress in research due to the erraticism of this disorder. Moreover, the study on developing tools or applications for parents, teachers, and healthcare workers to identify children who exhibit any form of autism is still a work in progress. The research work undertaken in this paper presents an analysis of supervised machine learning algorithms on mining interesting details that link the diverse nature of ASD and the possibility of computationally detecting markers for the syndrome. The preliminary findings on the performance of traditional machine learning algorithms in ASD classification is reported with the possibility of integrating deep learning architectures for ASD detection and therapy.