Behavioral Profiling of Children with ASD Using Score-Based K-Modes

Abstract

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that affects communication, social interaction, and behavior in children. This study analyzes children's behavioral profiles using K-Modes Clustering based on indication scores derived from seven ASD-related behavioral indicators. The dataset consists of 50 children's records, processed through data cleaning, missing-value imputation, indication-score formation, KModes clustering, and feature-relationship analysis using Cramer's V. The clustering produced three behavioral profile groups: Low Indication (22%), Medium Indication (44%), and High Indication (34%). Association analysis  shows the strongest relationship between speech ability and repetitive movements (Cramer's V = 0.548). These results indicate that K-Modes Clustering can serve as an initial exploratory approach for grouping observed behavioral indicator patterns, though the findings cannot yet be interpreted as diagnostic or clinical conclusions. This exploratory approach supports mapping the level of ASD-related behavioral indications to aid early, nonclinical screening in educational settings.