Clustering Language Learning Strategy Patterns in Children with Autism Using K-Prototypes

Abstract

Children with Autism Spectrum Disorder (ASD) have diverse language learning needs; however, appropriate learning strategies are often selected without considering individual characteristics. This study analyzes language learning strategy patterns for children with ASD using K-Prototypes Clustering. The dataset consisted of 57 records with features including age, gender, speaking ability, severity level, learning media, and learning difficulty. K-Prototypes was selected because it can handle numerical and categorical data simultaneously. The optimal number of clusters was determined using the elbow method based on the cost value. The results produced four clusters with 18, 14, 14, and 11 data points, respectively. Cluster stability assessed using the Adjusted Rand Index (ARI) showed an average value of 0.6838 (SD = 0.1734), indicating adequate stability. Chi-Square and Cramér's V tests showed significant associations between clusters and learning difficulty (V = 0.5496), speaking ability (V = 0.4406), and severity level (V = 0.4182), but not learning strategy (V = 0.1445, p = 0.1845). Learning strategy distributions also showed different dominant patterns across clusters. These findings provide an initial exploratory analysis of language learning strategy patterns among children with ASD.