📊 Student Performance Database

🔍
I am Trisa. Currently, I am working with a student performance database. I am planning to explore different kinds of relationships here. For example, observing the performance in grades vs study time is a potential one.
⚙️
It is easy to do a rendering of the complete data once. But to actively explore different relations, repeated rendering of the complete data is a computationally heavy operation.
🎯
What if I proceed with a sample? From the metadata, I know this dataset has around 10,150 entries. I can proceed with a sample of 1000 points (~10%), which should be fairly enough!
PerVis

Visualization Window

Marker Size: 2.0
References:
[1] Park, Y., et al., "Visualization-aware sampling for very large databases", ICDE 2016.
[2] Chen, Z., et al., "Variational blue noise sampling", IEEE TVCG, 18(10).
[3] Eldar, Y., et al., "The farthest point strategy for progressive image sampling", IEEE TIP, 6(9).
[4] Moumoulidou, Zafeiria, et al. "Perception-aware Sampling for Scatterplot Visualizations." arXiv preprint arXiv:2504.20369 (2025).

Controls & Steps

1 Initial Exploration
Generate a random sample to quickly build the analysis workflow.
👩🏽‍💻
Trisa
Data Scientist
Let's first understand my current scenario!