An intelligent assistant for mediation analysis in visual analytics
summary
A visual analytics prototype that runs automatic regression tests behind the scenes recommends three-variable plots, draws green arrows or flat bars as visual cues, and places key charts side-by-side for conflicting trends. It aims to detect when an apparent x–y correlation disappears after controlling for m and flags inconsistent mediation patterns. A five-participant exploratory study reports fewer wrong inferences and higher confidence, but results are preliminary and limited to three-variable regression checks.
AI-written summary · about
chapters (5)
- 0:00Start
- 0:20Intelligent Assistant for Mediation Analysis in Visual Analytics
- 3:05Pairwise Correlation
- 6:25Three Variable Visualization
- 7:40Conclusion
related talks
42mInteractive Data Analysis: Visualization and Beyond
39mVisual diagnostics for more informed machine learning
21mTurn Data into Action: How visualization aids decision intelligence
37mInteractive Explainable AI
22mOne-click analysis: Detecting and visualizing insights automatically
48mThe Agentic Data Science Multiverse
35mVisual Diagnostics for More Effective Machine Learning
31mFundamentals of Data Visualization