A COMPARATIVE STUDY OF MACHINE LEARNING MODELS FOR PREDICTION OF AUTISM SPECTRUM DISORDER USING SCREENING DATA

Yeap, Ming Yue (2023) A COMPARATIVE STUDY OF MACHINE LEARNING MODELS FOR PREDICTION OF AUTISM SPECTRUM DISORDER USING SCREENING DATA. [Final Year Project Report] (Unpublished)

	PDF Yeap Ming Yue (24 pgs).pdf Download (520kB)
	PDF (Please get the password by email to repository@unimas.my , or call ext: 082-583914/3973/3933) Ming Yue ft.pdf Restricted to Registered users only Download (2MB)

Abstract

Autism spectrum disorder (ASD) is a neurological and developmental disorder that affects how people interact with others, communicate, learn, and behave. ASD prediction is difficult because the diagnostic factors may not be based solely on observation. The project focuses on using ASD screening data to predict ASD traits in adults. This project aims to predict ASD traits in adults based on screening data using a machine learning approach. This can help them decide whether to seek a medical practitioner. The project proposed using classification, which is one of the machine learning approaches to predict autism spectrum disorder. The proposed prediction models are Logistic Regression, Random Forest, Support Vector Machine (SVM), K-Nearest Neighbours, Naïve Bayes, and Neural Network. The methodology adopted by the project is knowledge discovery in databases (KDD) to accomplish the needs of this project. The steps include domain understanding, data selection, data pre-processing, data transformation, data mining/modelling and model evaluation. The project will create a dataset based on AQ-10 adults questionnaire data that will facilitate future work in future work in predicting ASD in adults. Feature selection will be performed to find useful features in predicting ASD traits in adults. The performance of the classification models for ASD will be compared. Finally, the best classification model for ASD prediction was a model trained using the Support Vector Machine (SVM) algorithm

Actions (For repository members only: login required)

View Item