Adaptif Testlerde Öğrenci Yanıt Sürelerinin Zaman Serisi Kümelemesi: Akademik Başari İle İlişkilerinin İncelenmesi

Öz

Bu çalışma, bilgisayarlı uyarlanabilir testlerde öğrenci yanıt süresi kalıplarını analiz etmek ve akademik başarı ile ilişkilerini incelemek için zaman serisi kümeleme tekniklerini araştırmaktadır. Önerilen DETECT (Detection of Educational Trends Elicited by Clustering Time-series data) algoritması, zamansal dinamikleri davranışsal profillemeye entegre ederek klasik yöntemlerden ayrılır. DETECT, değişim noktası tespiti ve segment düzeyinde özellik çıkarımı ile gizli yanıt kalıplarını ortaya çıkarır ve sınav süreçlerine zamana duyarlı bir bakış sunar. 150 öğrencinin 30 maddelik matematik değerlendirmesinden elde edilen veriler, aykırı değerlerin çıkarılması ve z-skoru normalizasyonu ile ön işleme tabi tutulmuştur. DETECT, K-ortalamalar ve Hiyerarşik kümeleme yöntemleriyle karşılaştırılmıştır. ANOVA ve Tukey HSD testleri, K-ortalamalar ve Hiyerarşik kümeleme için anlamlı grup farklılıkları (p < 0.001), ancak DETECT için anlamlı olmayan sonuçlar (p = 0.1737) ortaya koymuştur. DETECT, yanıt süresi kalıplarını analiz ederek daha kişiselleştirilmiş ve tanısal açıdan zengin eğitim değerlendirmeleri için umut vadetmektedir

Anahtar Kelimeler

• Detect Algoritması
• K-ortalamalar algoritması
• Hiyerarşik kümeleme
• bilgisayarlı uyarlanabilir test

Time Series Clustering of Student Response Times in Adaptive Tests: Exploring Their Relationship with Academic Success

Öz

This study explores time-series clustering to analyze student response time patterns in computerized adaptive testing and their link to academic success. A novel algorithm, Proposed DETECT (Detection of Educational Trends Elicited by Clustering Time-series data), integrates temporal dynamics into behavioral profiling. Unlike traditional methods, DETECT applies change-point detection and segment-level feature extraction to reveal latent response patterns, offering a time-aware view of test-taking behaviors. Data from 150 students completing a 30-item mathematics assessment were preprocessed with outlier removal and z-score normalization. DETECT was compared to K-means and Hierarchical clustering. One-way ANOVA and Tukey HSD tests showed significant group differences for K-means and Hierarchical clustering (p < 0.001), but not for DETECT (p = 0.1737), highlighting its focus on temporal behavior over static performance. Correlation analysis found no significant link between average response time and scores. DETECT presents a promising tool for nuanced, personalized, and diagnostically rich educational assessments. Keywords: Detect Algorithm, K-means, Hierarchical clustering, computerized adaptive testing.

Anahtar Kelimeler

• Detect Algorithm
• K-means
• Hierarchical clustering
• computerized adaptive testing

Kaynakça

1. [1] S. L. Wise and X. Kong, "Response Time Effort: A New Measure of Examinee Motivation in Computer-Based Tests," Journal of Applied Measurement, vol. 6, no. 1, pp. 1-16, 2005.
2. [2] Tan, B. (2024). Response Time as a Predictor of Test Performance: Assessing the Value of Examinees' Response Time Profiles.
3. [3] De Boeck, P., & Jeon, M. (2019). An overview of models for response times and processes in cognitive tests. Frontiers in psychology, 10, 102.
4. [4] Araneda, S., Lee, D., Lewis, J., Sireci, S. G., Moon, J. A., Lehman, B., Arslan, B., & Keehner, M. (2022). Exploring Relationships among Test Takers’ Behaviors and Performance Using Response Process Data. Education Sciences, 12(2), 104. https://doi.org/10.3390/educsci12020104
5. [5] J. MacQueen, "Some methods for classification and analysis of multivariate observations," in Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability, vol. 1, 1967, pp. 281-297.
6. [6] S. C. Johnson, "Hierarchical clustering schemes," Psychometrika, vol. 32, no. 3, pp. 241-254, 1967.
7. [7] W. Li and J. Zhang, "A dynamic clustering approach for educational data with temporal dependencies," Journal of Educational Data Mining, vol. 8, no. 2, pp. 52-71, 2016.
8. [8] Y. Zheng and W. Yu, "Mining student behavioral patterns from clickstream data in online learning environments," Computers & Education, vol. 155, p. 103930, 2020.

9. [9] Dutt, A., Aghabozrgi, S., Ismail, M. A. B., & Mahroeian, H. (2015). Clustering algorithms applied in educational data mining. International Journal of Information and Electronics Engineering, 5(2), 112.
10. [10] Zhou, Y., & Paquette, L. (2024). Investigating Student Interest in a Minecraft Game-Based Learning Environment: A Changepoint Detection Analysis.
11. [11] Papamitsiou, Z., & Economides, A. A. (2014). Temporal learning analytics for adaptive assessment. Journal of Learning Analytics, 1(3), 165-168.
12. [12] JAlqahtani, A., Ali, M., Xie, X., & Jones, M. W. (2021). Deep time-series clustering: A review. Electronics, 10(23), 3001.
13. [13] S. Mao, C. Zhang, Y. Song, J. Wang, X.-J. Zeng, Z. Xu, and Q. Wen, “Time series analysis for education: Methods, applications, and future directions,” arXiv preprint arXiv:2408.13960, Aug. 2024.
14. [14] J. McBroom, K. Yacef, and I. Koprinska, “DETECT: A hierarchical clustering algorithm for behavioural trends in temporal educational data,” Journal of Educational Data Mining, vol. 14, no. 1, pp. 1–28, 2022.
15. [15] C. Romero and S. Ventura, “EDM 2.0: Temporal analytics in learning systems,” Education and Information Technologies, vol. 27, no. 8, pp. 11471–11500, 2022.
16. [16] J. Paparrizos, F. Yang, and H. Li, “Bridging the gap: A decade review of time-series clustering methods,” arXiv preprint arXiv:2412.20582, Dec. 2024.
17. [17] E. Anghel, L. Khorramdel, and M. von Davier, “The use of process data in large-scale assessments: a time-series approach,” Large-Scale Assessments in Education, vol. 12, art. 13, 2024.
18. [18] F. Cobo, I. Koprinska, and Y. Rogers, “A strategy based on time series and agglomerative hierarchical clustering to model forum activity in e-learning,” in Proc. 4th Int. Conf. on Educational Data Mining, Eindhoven, Netherlands, 2011, pp. 21–30.
19. [19] M. Malik, H. Zhou, and L. Sun, “Dynamic feature ensemble evolution for enhanced feature selection,” Scientific Reports, vol. 15, no. 2, pp. 112–126, 2025.
20. [20] F. Chang, Y. Ji, L. Liu, Y. Xiao, B. Chen, and H. Liu, “Analysis of university students’ behavior based on a fusion k-means clustering algorithm,” Applied Sciences, vol. 10, no. 18, art. 6566, 2020.
21. [21] T. Mai, M. Bezbradica, and M. Crane, “Learning behaviours data in programming education: community analysis and outcome prediction,” Future Generation Computer Systems, vol. 127, pp. 42–55, 2022.
22. [22] E. Iatrellis, I. K. Savvas, P. Fitsilis, and V. C. Gerogiannis, “Temporal feature fusion for predicting student outcomes,” Expert Systems with Applications, vol. 213, p. 119284, 2023.
23. [23] A. Hung, B. Wu, and T. Wu, “Early warning via time-series clustering in adaptive learning systems,” IEEE Transactions on Learning Technologies, vol. 15, no. 1, pp. 45–55, 2022.
24. [24] W. Xing, D. Du, and L. Ma, “Change point detection for learning behaviors in time-series data,” Journal of Educational Data Mining, vol. 15, no. 1, pp. 1–30, 2023.
25. [25] V. Kovanović, D. Gašević, S. Dawson, and G. Siemens, “Dynamic time warping for educational sequence alignment: Applications and benchmarks,” Computers & Education, vol. 180, p. 104432, 2022.
26. [26] Q. Liu and S. Tong, “Educational data mining: A 10‑year review,” Discover Computing, vol. 3, no. 1, pp. 1–24, 2025.
27. [27] F. Goldhammer, J. Naumann, and I. Stelzl, “Process data in large scale computer based assessments: Insights into students’ strategies and processes,” Educational Measurement: Issues and Practice, vol. 36, no. 4, pp. 16–29, 2017.
28. [28] W. Xing and D. Du, “Detecting change points in student learning processes with sequential data analysis,” Journal of Educational Data Mining, vol. 11, no. 1, pp. 1–24, 2019.
29. [29] T. Shimizu, Y. Tanaka, and K. Watanabe, “Uncovering classroom behavioral shifts via PELT-based change point detection,” Smart Learning Environments, vol. 11, no. 2, article 7, 2024
30. [30] Y. Lee and J. Jia, “Using response times to identify rapid guessing behavior in computer based tests,” Educational and Psychological Measurement, vol. 74, no. 5, pp. 785–802, 2014.
31. [31] G. J. Ryzin, “Identifying learning states and behavioral profiles with response time data,” Psychometrika, vol. 83, no. 2, pp. 481–507, 2018.
32. [32] R. S. Baker and P. S. Inventado, “Educational data mining and learning analytics: An introduction,” in Learning Analytics and Knowledge, Springer, 2014, pp. 64–75.
33. [33] R. Fernández Alonso, J. A. Suárez Álvarez, and R. Muñiz, “Imputation methods for missing data in educational diagnostic evaluation,” Electronic Journal of Research in Educational Psychology, vol. 9, no. 3, pp. 1033–1052, 2011.
34. [34] M. elmakkaoui, “Applying Z score Normalization on Time Series Data for a Machine Learning Model,” Medium, Jun. 16, 2025.
35. [35] M. R. Berthold and F. Höppner, “On clustering time series using Euclidean distance and Pearson correlation,” arXiv:1601.02213, 2016
36. [36] W. Wang, S. Xu, and Q. Li, “Uncovering student problem solving strategies using eye tracking and response time data in an interactive simulation,” Journal of Educational Psychology, vol. 115, no. 1, pp. 1–17, 2023.
37. [37] M. K. Alian, H. M. Alian, and I. Z. M. Alian, “Clustering Student Sequential Trajectories Using Dynamic Time Warping,” in Proc. Int. Conf. Educational Data Mining, 2020, pp. 1–10.
38. [38] H. S. Lim, B. J. Kim, and S. Y. Choi, “Time Series Analysis of VLE Activity Data,” in Proc. 9th Int. Conf. EDM, 2016, pp. 1–8.
39. [39] J. Sun and Q. Lu, “A review of using response time in educational assessment,” Frontiers in Psychology, vol. 12, article 683401, 2021.
40. [40] Qualtrics, “What is ANOVA (Analysis Of Variance) and What Can I Use It For?,” Qualtrics.com, 2024.
41. [41] DataScientest, “Pearson and Spearman Correlations: A Guide to Understanding and Applying Correlation Methods,” DataScientest.com, Jan. 19, 2024.
42. [42] Nakamura, K., Ishihara, M., Horikoshi, I. et al. Uncovering insights from big data: change point detection of classroom engagement. Smart Learn. Environ. 11, 31 (2024). https://doi.org/10.1186/s40561-024-00317-6
43. [43] Shen, D. S., & Chi, M. (2023). TC-DTW: Accelerating multivariate dynamic time warping through triangle inequality and point clustering. Information Sciences, 621, 611-626.
44. [44] Zhang, H. (2025). AI-driven innovation and entrepreneurship education: K-means clustering approach for Chinese university students. Discover Artificial Intelligence, 5(1)
45. [45] Hao, Z., Jiang, J., Yu, J., Liu, Z., & Zhang, Y. (2025). Student engagement in collaborative learning with AI agents in an LLM-empowered learning environment: A cluster analysis. arXiv preprint arXiv:2503.01694.
46. [46] Heikkinen, S., Saqr, M., Malmberg, J., & Tedre, M. (2025). A longitudinal study of interplay between student engagement and self-regulation. International Journal of Educational Technology in Higher Education, 22(1), 21.