An Exploratory Data Analysis of Agentic System Use on the Chaoxing Platform: Behavioral Correlates, Purposes, and Perceptions
Students increasingly use AI-powered features on learning platforms. We explore AI companion usage patterns on the Chaoxing (Superstar) Learning Platform using survey data from 75 university students. Using a three-tier EDA framework (univariate description, bivariate correlation, and multivariate exploration), we examine usage behaviors (frequency, duration, depth), usage purposes (seven categories), demographic differences, and user perceptions. Key findings reveal a significant positive correlation between usage duration and depth (Spearman ρ = 0.325, p < 0.01), suggesting a learning curve effect. Usage purposes are significantly associated with behavioral depth: students using AI for assignments (p = 0.005), emotional support (p = 0.011), and AI Q&A (p = 0.023) demonstrate deeper engagement. The strongest purpose co-occurrence is between assignments and instructor requirements (Jaccard = 0.672). In addition, demographic variables (gender, grade, major) show no significant association with usage behavior, with similar usage patterns across groups. User perception analysis reveals ‘convenient’ as the dominant keyword (n = 22), surpassing ‘useful’ (n = 10). Overall, usage correlates with depth, purposes differentiate depth, and convenience dominates perceptions; findings are relevant to platform design.
[1] Dwivedi YK, Kshetri N, Hughes L, et al., 2023, Opinion Paper: “So What If ChatGPT Wrote It?” Multidisciplinary Perspectives on Opportunities, Challenges and Implications of Generative Conversational AI for Research, Practice and Policy. International Journal of Information Management, 71: 102642.
[2] Bai Y, Wang S, 2025, Impact of Generative AI Interaction and Output Quality on University Students’ Learning Outcomes: A Technology-Mediated and Motivation-Driven Approach. Scientific Reports, 15(1): 24054.
[3] Kasneci E, Sessler K, Küchemann S, et al., 2023, ChatGPT for Good? On Opportunities and Challenges of Large Language Models for Education. Learning and Individual Differences, 103: 102274.
[4] Cengiz S, Peker A, 2025, Generative Artificial Intelligence Acceptance and Artificial Intelligence Anxiety Among University Students: The Sequential Mediating Role of Attitudes Toward Artificial Intelligence and Literacy. Current Psychology, 44(9): 7991–8000.
[5] Ursavaş Ö, Yalçın Y, İslamoğlu H, et al., 2025, Rethinking the Importance of Social Norms in Generative AI Adoption: Investigating the Acceptance and Use of Generative AI Among Higher Education Students. International Journal of Educational Technology in Higher Education, 22(1): 38.
[6] Morgenthaler S, 2009, Exploratory Data Analysis. WIREs Computational Statistics, 1(1): 33–44.
[7] Tian X, Ayanwale M, Molefi R, et al., 2026, Decoding Symmetric and Asymmetric Pathways in Generative AI Learning Adoption: A Multi-Method Study. International Journal of Educational Technology in Higher Education, 23(1): 37.
[8] Isaeva R, Caner H, Caner M, et al., 2026, Students’ Engagement with Generative AI in Academic Learning: A Self-Determination Theory and Epistemic Network Analysis Study. Computers and Education: Artificial Intelligence, 10: 100606.
[9] Li J, Zhang J, Chai CS, et al., 2025, Analyzing the Network Structure of Students’ Motivation to Learn AI: A Self-Determination Theory Perspective. npj Science of Learning, 10(1): 48.
[10] Güngören Ö, Gür Erdoğan D, Horzum M, 2026, University Students’ Acceptance of Generative Artificial Intelligence Tools: A Mixed-Methods Study on Opinions, Attitudes, and Behavioral Intentions. BMC Psychology, 14(1): 238.
[11] Belawati T, Prasetyo D, 2025, Generative AI-Based Tutoring for Enhancing Learning Engagement and Achievement. Open Praxis, 17(2): 211–226.
[12] Ravšelj D, Keržič D, Tomaževič N, et al., 2025, Higher Education Students’ Perceptions of ChatGPT: A Global Study of Early Reactions. PLoS ONE, 20(2): e0315011.
