Data Visualization
By Shaun Patel at Tue, Nov 17 2009 3:34pm |
I am looking at creating new and innovative ways of displaying multi-terabytes of electrophysiology data. This involves developing methods for displaying data, as traditional 2-d plots simply don't allow us to see the complete result.
In my research, I have run into an interesting flash based method called Flare (http://flare.prefuse.org/) for potentially displaying data in various formats interactively. I was wondering if anyone has experience with this for scientific data sets. Secondarily, I am not too familiar with programming in flash; is it possible to wrap matlab programs (for computing and analyzing the data) around a flash program (for displaying the data)?
Cheers.
references on data visualization and related topics
Howdy, Shaun,
Flare looks interesting. I haven't used it, though. Since flare.data.DataSource class makes it possible to access data from a URL, perhaps you could devise a way to share data between Flare and Matlab without having to bundle one inside the other.
There are lots of packages to check out here
http://homepage.mac.com/swain/Macinchem/Static/data_anal_tools.html
originally posted in this useful thread:
http://www.macresearch.org/software-data-visualization
From your post I'm not certain how much experience you may have in data visualization--perhaps much more than I--or how much you may have read on the subject. Since you're looking at "new and innovative ways" I thought I'd create a list of references. One or more of these sources could help someone concerned with data visualization to identify a visualization method and choose a software package, or perhaps to inspire the creation of a data visualization technique most suitable for a particular field.
1. Information Visualization: Beyond the Horizon by Chaomei Chen
Covers a variety of data visualization techniques including hyperbolic views, treemaps, etc.
2. Information Visualization: Perception for Design by Colin Ware
Perception and data visualization. Starts with human visual perception, covers data visualization techniques.
3. The Humane Interface by Jef Raskin
An introduction to zooming user interfaces (ZUIs). This is a relatively quick read, and it puts a number of topics into perspective: HCI, data visualization, ease of use, etc. Other books on HCI, user interaction, and user design go into more detail on implementation and human factors. If you're not already familiar with ZUIs, then I strongly recommend reading this book, especially since you're dealing with large data sets.
4. The Grammar of Graphics by Leland Wilkinson
From the back cover: "...presents a unique foundation for producing almost every quantitative graphic found in scientific journals, newspapers, statistics packages, and data visualization systems." The book is informative, well organized, thought-provoking, and even entertaining.
5. The Visual Display of Quantitative Information by Edward Tufte (+ 3 more books)
An elegant guide to designing graphics. Required reading for anyone involved in data visualization.
6. Universal Principles of Design by Lidwell, Holden, and Butler
Bite-sized topics in design. This book is another good source of inspiration, especially if you need not only to present information but also design the user interaction.
Good luck.
---
"A life lived alone weakens the mind's immune system, and your brain becomes susceptible to an attack of strange ideas." - A Fraction of the Whole, Steve Toltz
Rethunk, Many thanks for
Rethunk,
Many thanks for the excellent post. I will most definitely look into these books- I had no idea there is such an extensive literature base on this field.
Unfortunately, I am not that well-versed in complex methods and techniques for visualizing data. All of this, for me, stemmed from a talk that I saw on Ted.com (http://www.ted.com/talks/hans_rosling_reveals_new_insights_on_poverty.html). Although the topic is completely unrelated to my area of interest, it occurred to me the potential for missing very powerful and interesting phenomena within very large data sets, when using inadequate methods for examination.
In terms of electrophysiology, the data analyses still are fairly primitive with a few exceptions here and there. With increasing technological development, scientists have been able to increase the number of channels of data acquisition exponentially, with very limited growth in the mathematical/computational side of analyzing and visualizing this data.
Hence my previous post. My intentions were to try to 'ping' the audience for interesting ideas to develop new methods for studying the brain with data we already have. Thanks for the help!
Cheers,
Shaun
Scagnostics
Shaun,
You're certainly welcome.
I watched the whole Hans Gosling video; that's quite a nice stats/graphics tool he's using.
The number of books on data visualization has increased in recent years, and if you dig around you'll find a number of older books that are useful, too. Tufte makes several references to Tukey.
When displaying and analyzing data it can sometimes help to expand the range of tools beyond the common statistical methods, too. You might want to look into "scagnostics" (scatterplot diagnostics), which could help you to analyze and/or categorize your data using image analysis algorithms. The idea originally proposed by Tukey appears to be spreading. (For more about image analysis algorithms, check out the entries for "computer vision" and "machine vision" on Wikipedia and other sites.)