Analyzing Nursing Students’ Learning Behaviors and Outcomes Using Data-Mining Techniques: A Case Study of the Fundamentals of Nursing Course

The study’s data are systematically collected, preprocessed, and analyzed to explore the link between nursing students’ learning behaviors and academic performance, with details as follows:

1. Data Sources & Participants
The primary data source is the Superstar Learning management system (LMS), which provides real-time, complete behavioral and academic data. Participants are 323 undergraduate nursing students from six classes of the 2023 cohort (77 males, 23.85%; 246 females, 76.16%; aged 18–23 years).

2. Data Categories & Key Indicators
Two core data categories are collected across two semesters:
Learning behavior data: Login frequency, time-on-task, video viewing rate, resource-click paths, interactive discussion participation rate, assignment submission rate, forum posting rate, and extended resource click-through rate. A subset of task completion data (Figure 2) shows, for example, 293 students completed platform login (Task 1) vs. 31 who did not, and 304 completed assignment submission (Task 5) vs. 20 who did not.
Academic performance data: Chapter-quiz attempt frequency, stage-test scores (across 4 stages: Chapters 1–5, 6–10, 11–13, 14–18), and final examination scores.

3. Data Preprocessing
After collection, data undergo validation (to ensure completeness/consistency), outlier handling (removal or replacement), and structuring (converting unstructured text to structured format) before storage in a database—laying the groundwork for subsequent analysis.

4. Data Analysis & Key Results
Clustering analysis (k-means, SPSS 25.0): With k=4, students are grouped into 4 clusters (Figure 4):
Cluster A (n=60, High-Frequency Interaction): Highest test score (97.73) and forum posting rate (95%).
Cluster B (n=127, Resource-Oriented): High video viewing rate (99.17%, comparable to Cluster A) and test score (85.90).
Cluster C (n=103, Passive Participation): Moderate test score (72.23) and low assignment submission rate (76.90%, similar to Cluster D).
Cluster D (n=33, Low-Activity): Lowest test score (63.32) and video viewing rate (<60%).

Intervention & outcome data: Post-personalized interventions, Cluster D’s video viewing and assignment submission rates improved; all clusters showed score growth (Figure 5). Second-semester final scores (80.42±7.40) were significantly higher than the first (76.02±8.67, t=23.88, P<0.001). Inter-cluster differences in chapter-quiz frequency (Table 1, e.g., Chapter 3: 2.98±0.79 for A vs. 6.00±1.06 for D, F=72.03, P<0.01) and test scores (Table 2, Stage 1: 90.02±1.08 for A vs. 59.58±1.48 for D, F=4759.47, P<0.001) were statistically significant.