Lecture 2

Question 1

Three different stages to multivariate visualisation pipeline

Answer 1

Data transformation, visual mapping, view transformation

Question 2

Data transformation

Answer 2

Computation-centric and focus on quantitative results

Question 3

Dimension reduction

• What does it used for?
• How is the scalability of it is addressed?

Answer 3

• Capturing the overall structure of the dataset.
• Control point based projection approaches and approximations.

Question 4

Subspace clustering

• What is the motivation
• What does it do?

Answer 4

• A single embedding is unlikely to be sufficient for understanding the data.
• Identify multiple informative 2d projections. Either finds clusters in subset of dimensions, or cluster points that share a low-dimensional linear subspace.

Question 5

Regression analysis

Answer 5

Provides a class of methods designed to capture the quantitative relationship among individual dimensions. More effective parameter tuning and exploration

Question 6

Visual mapping

What are the main approaches?

Answer 6

Different arrangements of data axes, glyph-based, pixel-oriented, hierarchy-based

Question 7

Data axes based visual mapping

• What is it?
• What is the challenge?
• How to tackle this challenge?/

Answer 7

• Different arrangements of axes.
• number of possible rankings increases as dimensionality increases.
• Quality metrics for filtering the interesting configurations.