Visualising Complex Data - PowerPoint PPT Presentation

1 / 14
About This Presentation
Title:

Visualising Complex Data

Description:

Find the underlying meaning in the data. Use: Compression techniques. Dimensionality reduction. ... Data points linked together by springs - Hooke's Law. ... – PowerPoint PPT presentation

Number of Views:28
Avg rating:3.0/5.0
Slides: 15
Provided by: vintechn
Category:

less

Transcript and Presenter's Notes

Title: Visualising Complex Data


1
Visualising Complex Data
  • A force based model for visualisation

2
Data Analysis
  • What does this data represent?
  • Find the underlying meaning in the data.
  • Use
  • Compression techniques.
  • Dimensionality reduction.

3
Complex Data
  • Complex data
  • High number of dimensions.
  • Intricate interdependencies.
  • Difficulty in applying traditional techniques.
  • Leads to misinterpretation of results.

4
Visualising Complex Data
  • Our brains are good at visually perceiving
    patterns.
  • This is useful when interpreting complex data
    through visualisation.

5
Force Based Visualisation
  • Cercia has developed tools for cluster, network
    and process visualisation.
  • Dimensionality reduction.
  • Good for small/medium data sets.
  • Highly flexible (networks, clusters, etc.)?
  • Retains symmetry.
  • Intuitive and interactive

6
Forces
  • Randomly place each data point in 3D space.
  • Apply forces
  • All data points repel each other like
    electrically charged particles - Coulombs Law.
  • Data points linked together by springs - Hookes
    Law.
  • Spring strength depends on similarity between
    data points.

7
Crime Data
  • Crime data provides a powerful tool
  • Detection and prevention.
  • Tracking individuals.
  • Investigating specific crimes.

8
Crime Networks
  • Identify specific criminal networks.
  • The relationship between
  • Individuals
  • Resources (cars, addresses, etc.).
  • Identify correlations within the network.
  • Determine best means of disruption.

9
Analysis of Crime Networks
  • Manual analysis
  • Slow, difficult and tedious.
  • Data can be, complex, high dimensional,
    disparate.
  • Automated Analysis
  • Retrieve information about specific instances.
  • Can identify specific instances.
  • Hard to get an overview!

10
Demonstration
11
Visualising Criminal Networks
  • Nodes Individual criminals
  • Links Specific crimes

12
Merging Vehicle Data
  • Merge individual vehicles into crime network.

13
Merge Postal Code Data
  • Criminals and their crimes can be associated with
    specific localities.

14
Applications
  • Processing credit applications
  • Detecting and classifying fraud
  • Web log analysis
  • Analysis of social networks
  • Doesnt require you to know what your looking
    for!
Write a Comment
User Comments (0)
About PowerShow.com