Information Technology: Communication

1
Information Technology Communication Pedagogy

• Deborah Nolan, UC Berkeley
• Duncan Temple Lang, UC Davis

2
What is Information Technology?

• Technology embodies the human knowledge of
  solving real problems
• The use of computers and software to
• Design
• Collect/access
• Clean
• Process / model fit
• Report findings
• including text, images, sound

3
Communication Pedagogy

• Changing what we teach our students
• Facility with new data technologies
• Changing how we teach
• Balance between math and computer
• Beyond the applet
• Changing how we communicate results
• Dynamic, interactive, reproducible research

4
Why include IT in the Stat program?

• Experience in problem solving
• Why teach data analysis?
• Facility with technology
• Enable more interesting/meaningful collaborations
  
• Adapt to new technologies
• Conduit that brings us data and through which we
  disseminate results and tools
• IT as a source of data
• Sensor nets, Intrusion detection, Internet packets

5
IT in the Statistics Program

• Design - Simulation studies
• Collect/access
• Databases, XML, SOAP, HTTP
• Clean - Regular expressions, statistical
  anomolies
• Analyze/process
• High dimensional model, model validation,
  optimization
• Report
• Visualization, reproducible research

6
Statistics Programming

• Communication software before publications
• Ad hoc approach
• Judgement
• Limitations
• Software medium of communication
• Structure and style extensible
• Project organization CVS, shell tools

7
Changing how we teach

• Math rigor and beauty
• Math easy and natural for us
• Math tool for succinct expression
• Computing
• Different/complementary communication paradigm
• Technology is lacking for easy use
• Scientific application
• Methods in context better understanding
• Statistical, exploratory, and critical thinking

8
Our Approach

• Guided investigations in context
• Adapt feedback dynamically to reader input
• Encourage what-if investigations
• Eliminate black-box for authors
• Authors easily modify content
• Allow readers to drill down
• Extensible to different audiences

9
Our Approach

• Tailor to context/application
• GUIs Specialized plots
• Multiple approaches
• Spreadsheets visual computations
• Functions automated computations
• Easy entrance for readers
• Sliders and buttons vs Command line
• Supports variety of levels
• Executive vs spcifics of implementation

10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Technology

• XML
• Pervasive (office, gnumeric)
• Extensible
• Self describing structured data/text
• Offers different views
• XSL has an embedded R interpreter
• XML is embedded in R

14
Technology

• Gtk
• Offers suite of tools, e.g. spreadsheet, word
  processor, browser
• Supported on many platforms
• Extensive GUI Toolkit
• Large community of developers

15
Technology

• R language
• Access existing statistical functionality
• Well tested code
• Bindings to Gtk Toolkit, DCOM, Perl
• Large community of users researchers

16
(No Transcript)
17
(No Transcript)
18
Infrastructure - Interfaces

• RGtk - GUIs integrated with R
• RGtkHTML Our own widget to display HTML
• win lt- gtkWindow()
• h lt- gtkHTML()
• hAddCallback(link_clicked,
• function(w, link) ... )

19
Infrastructure - Interfaces

• XML XML parser
• Dynamic and interactive content
• ltOBJECT typeapp/x-R-device WIDTH400gt
• ltparam nameinit valueplot(g)/gt
• lt/OBJECTgt
• ltfeedbackgt
• paste(.totalScore, points accumulated)
• lt/feedbackgt

20
Infrastructure - Interfaces

• Rgnumeric R plug-in for Spreadsheet
• Ggobi Gtk based dynamic plots
• RDCOM Interface with Excel, Word
• SNetscape R Plugin to browser/Javascript
• http//www.omegahat.org

21
Spreadsheet
R
Data output
GUI controls
22
Changing communication of results

• Communication
• Students tutorial
• Researchers new methodology
• Applications user interface
• Redefine the Document
• Interactive constrained interactive programming
• Dynamic flexible representation
• Reproducible contains computations, not just
  output

23
Document ModelViewController

• Model
• Encapsulate text, data, functions, and how a
  process works
• View
• Renders content to a display
• Multiple views / Composite with sub-views
• Controller
• Instructs model and view to perform based on
  reader input

24
XML code, data, text
R
New Display
25
Google Page Ranking - Model

• How does Google determine the ranking of the
  pages it returns from a query?
• Connectivity matrix -
• Random walk on the web
• Markov chain Transition probabilities
• Markov chains
• simulate chain to explore properties
• stationary distribution

26
Browser
Browser
Grid Trans Prob
Graphics Network

Link - new page
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)