Implementation of The Random Forest Algorithm for Early Detection Indications of Autism in Special Needs School (SLB) Students

Bagus Tri Mahardika (Darma Persada University)
Duha Nur Pambudi

Journal Technology Information and Data Analytic
Vol 3 No 1 (2026): Journal Technology Information and Data Analytic

This study aims to develop a system for early detection signs of autism in pupils at Special Needs Schools (SLB) by applying the Random Forest method. The problem addressed is how to provide an accurate and easily accessible tool for the early identification of signs of autism. The solution involves developing a Random Forest-based classification model using data from the Autism Spectrum Quotient (AQ-10) questionnaire, and then integrating it into a web application system built with a PHP frontend and a Flask backend. This system allows users to complete the questionnaire, upload data, and obtain prediction results automatically. Test results show that the model has an average accuracy of 99%, precision of 98%, recall of 100%, and an F1-score of 99%, as well as an AUC value above 0.98 in every fold. Consequently, this system is effective as a tool for initial screening to detect signs of autism in students at special schools in a practical and efficient manner.