The Prediction and Investigation of Factors in the Adaptability Level of Online Learning Based on AutoML and K-Means Algorithm

This study applies advanced Artificial Intelligence (AI) tool to analyze 1,205 students, aiming to enhance adaptability prediction and identify key influencing factors. Using K-means clustering, students were categorized into two groups, revealing distinct learning behaviors and characteristics. Automated Machine Learning (AutoML) technology, utilizing Baidu's EasyDL, predicted online learning adaptability for each group, which is assessed by various performance indicators. The results demonstrated that the accuracy of adaptability prediction for the two groups, Cluster_0 and Cluster_1, was 91.3% and 98.1%, respectively. The performance indicator analysis reveals the superiority of Cluster_1 in predicting high adaptability levels, while Cluster_0 performed poorly in the high-high category. The importance scores of key factors further reveal the multidimensional factors affecting students' adaptability to online learning. For instance, in Cluster_0, the top three key factors are Location, IT Student, and Load-shedding, while in Cluster_1, they are Internet Type, Device, and Financial Condition. An in-depth comparative analysis of the factors in both groups highlighted the significant impact of students' gender and age on their adaptability levels. Therefore, in the practical application of the predictive model, personalized analysis and intervention should be conducted according to the specific characteristics of different groups.