Perbandingan Model Machine Learning untuk Klasifikasi Autism Spectrum Disorder pada Toddler Screening Dataset

Abstract

 

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by impairments in social communication, social interaction, and repetitive behaviors. Early detection of ASD in toddlers is important to support faster and more appropriate intervention. Previous studies have applied machine learning algorithms, such as Support Vector Machine (SVM), Random Forest, Naive Bayes, and K-Nearest Neighbor (KNN), for ASD classification with relatively high accuracy. However, variations in datasets, data imbalance, and algorithm selection still affect classification performance. Therefore, this study proposes the implementation of several machine learning algorithms for ASD classification using the ASD Toddler Screening Dataset. A comparative approach was used to evaluate the performance of Decision Tree, Random Forest, Logistic Regression, KNN, Naive Bayes, and SVM algorithms. The research stages included data preprocessing, categorical attribute transformation using encoding and one-hot encoding, exploratory data analysis (EDA), data splitting with an 80:20 train-test ratio, and handling imbalanced data using the Synthetic Minority Over-sampling Technique (SMOTE). Model evaluation was conducted using accuracy, precision, recall, and F1-score metrics. The results showed that SVM achieved the best performance with an accuracy of 100%, followed by Logistic Regression (98.58%), Random Forest (95.73%), Decision Tree (93.36%), KNN (92.89%), and Naive Bayes (87.20%). The findings indicate that SVM provides superior performance compared to other machine learning algorithms in classifying ASD in toddlers.