windows programming parallel coordinates adithya addanki(aa207) instructor: dr. yingcai xiao
TRANSCRIPT
2
• To know– What– Why
• Implementation Details• Technical Challenges• Demo• Limitations• Applications• Extensions
Agenda
3
What and Why
• Most of the real world data typically has >=3 dimensions• Curse of Dimensionality
• Understanding data of more than 3 dimensions• Reduction techniques might be applied
• Backdrop of these might be the user want to still analyze the data based on original attributes instead of transformed or reduced attributes.
• Visualization• Visualizing dimensions >3 is not perceivable by human eye.
• Parallel Coordinates• Philbert Maurice D’Ocagne in 1885• Representing multiple dimensions with the help of parallel axes.• Samples in data set are represented as poly lines spanning the axes.
7
Implementation Details
• WPF– UI
• Code Behind- Application Logic– C#
Challenges• Representing heterogeneous data types• Scaling the range of values to fit to fixed axes
Demo
8
Limitations
• Each axis can only be compared with two other axes at once;– One to the left another to the right
• More range of values in each attribute might cause confusion to the user. Use with caution with regards to categorical data.
• ASCII based text characters only[character sets and digits]• Rearrangement of attributes | axes order might be required to
view patterns– Good arrangement of axes could be dealt using heuristics.
9
Extensions & Applications
• Brushing- A provision to analyze the parallel coordinate plot where clusters are formed.
• Parallel Coordinates Applications– Data Mining– Traffic Control– Process Control– Computer Vision