Performance Analysis of Clustering Algorithms for Dyslexia Detection

Abstract

Clustering algorithms plays a vital role in analyzing and evaluating vast number of high dimensional health care data ranging from medical data repositories, clinical data, electronic health records, body sensor networks, IoT devices, and so on. Dyslexia, a learning disorder, is a common problem that is found in children during the initial stages of formal education, which is detected as mild to severe. It can also be one of the reasons of failure in the school. According to the literature, this difficulty is commonly seen among special needs children in education. There are few studies focused on the application of classification algorithms for detecting the presence of dyslexia. This paper focuses one of the SDG goals, Goal 4: Quality Education, as dyslexic students can be given equal and quality education. Analyses of an online gamified test-based dataset is done by applying various clustering techniques, such as K-Means, Fuzzy C-Means, and Bat K-Means to assess their effectiveness in detecting the problem dyslexia. As the dataset is large, it is observed that usage of clustering methods gives us gain insight into the distribution of data to observe characteristics of each cluster. The clustering results are evaluated using root mean squared error (RMSE), mean absolute error (MAE), Xie-Beni index, and it is found K-Means outperforms FCM and Bat K-Means algorithm for analyzing different levels of the learning disorder.