Structuring Interactive Cluster Analysis

Table of Contents

Structuring Interactive Cluster Analysis

Structuring Interactive Cluster Analysis

Overview - argument

Overview - contents

Problem - visual groupings

Problem - geometry exploited

Problem - context matters

Problem - close documents

Problem - distance as angle?

Problem - MRI brain segmentation

Problem - MRI brain aneurysm

Problem - MRI mult-channel

Problem - biology and abstract groups

Problem - ill defined

Computational resources

Computational resources (and response) - processing

Computational resources (and response) - memory

Computational resources (and response) - display

Computational resources - balance

High interaction (much overlooked by researchers)

Example: image analysis … find groups via intensity (contours and two small unusual structures revealed)

Example: image analysis … other measurements may contain interesting structure

Example: image analysis … identify new structure location in the original image

Example: image analysis … mark new groups by colour (hue, preserving lightness in original image)

Example: image analysis … explore relation between old and new groups via contours in the image itself

Example: 8 dimensions from teeth measurements on species (+ sex)

Example: apes, hominids, modern humans

Example: mutual support and shapes

Example: mutual support and shapes

Example: mutual support and shapes

Example: mutual support and shapes

Example: exploratory data analysis

Example: exploratory data analysis

Example: exploratory data analysis

Example: exploratory data analysis

Interactive clustering

Automated clustering: typical software

Example: K-means clustering

Example: K-means clustering

Example: VERI  Visual Empirical Regions of Influence

Example: VERI  Visual Empirical Regions of Influence

Visual Empirical Regions of Influence

Example: VERI

Example: VERI (with parameters)

Integrating automatic methods:

Refine

Reassign

Refinement sequence:

Refinement sequence:

Refinement sequence:

Refinement sequence:

Refinement sequence:

Reassign, reduce sequence:

Explore present partition:

Partition to be reduced:

Reduce sequence:

Reduce sequence:

Reduce sequence:

Moves (generic functions)

Challenges:

A prototype interface

Interface illustration: details of moves

Interface - reduce

Interface - refine

Interface - reassign

Interface illustration: example of use

Interaction

Interaction

name and save partition

prototype - refine to 4

prototype - refine to 5

Select nested partitionsand view dendrogram

Reassign, dendrogram updated

Cluster plot + dendrograminteraction movie

Other operators

Creation:

Composition:

Implications:

New problems:

Summary

Related references:

Acknowledgements:

Email: rwoldford@uwaterloo.ca

Home Page: http://www.stats.uwaterloo.ca/~rwoldfor

Other information:
This is an annotated and extended version of a presentation first given at the Statistical Society of Canada on June 9, 2003 in Halifax, Nova Scotia. Another version was given at the Department of Computer Science of the Memorial University of Nfld.

Download presentation source