Autism spectrum disorder (ASD) is a complicated neurological illness characterised by impairments in social interaction, communication, and restricted behaviour. In order to maximise therapy and improve long-term outcomes, early and accurate diagnosis of ASD is crucial. Using machine learning techniques in the diagnostic process has shown promise in efficiently evaluating large datasets and discovering significant trends. The ability to improve prediction accuracy and resilience via the use of ensemble techniques has attracted a lot of attention. This paper offers a thorough analysis of the existing research on the topic of machine learning methods for ASD prediction. It describes the research dataset and the steps required to prepare it for analysis. Also elaborated upon are the ensemble methods used for this investigation. The results and evaluation of the models' efficacy are presented in the study. In addition, the study's limitations are discussed, as are the difficulties faced. These findings highlight the promise of ensemble methods in improving the precision of ASD prediction and hence, their potential contribution to early intervention efforts.