SA Progress Report: June 2013

Interface

The current focus of this project is to develop a graphical user interface that will allow users to easily convey all of their substantive understanding of a dataset so as to allow automated robust model selection.  

Towards that end, recent research this month has involved:

• Building a user interface in D3 (Javascript) to compose directed graphs of relationships between variables.
• Exploring current methods to call R processes from Javascript, including RServe.
• Adding Bayes Factors and Out-of-Sample Forecasting to models in Zelig to evaluate model selection. 

The forthcoming timetable for this work includes:

• On August 13 we are demonstrating our design at an evaluation meeting. 
• On August 16 we are submitting a proposal for grant funding.

On July 1, Vito D'Orazio began work on Zelig as a Postdoctoral Researcher.  Vito has experience in machine learning algorithms, especially for large scale automated data collection and coding, such as the forthcoming 4th release of the Militarized Interstate Dispute (MID) dataset.