The Ringtail Computational Platform for Economics
The Ringtail project is a research data system that supports a range of novel data-intensive economics applications. In particular, it focuses on processing and querying social media data. It includes:
- Large-scale archiving of social media streams
- Query tools for quickly exploring social media time series
- Autosuggest-driven tools that enable domain experts to rapidly formulate domain-specific time series indexes
The query workload and research challenges for Ringtail are driven by our collaboration with practicing economists. You can see the
first results of our collaboration here.
We have created two Ringtail-based systems so far. The first is a site for predicting unemployment insurance claims, updated weekly. The second is a slightly-obsolete video of our query tool system (formerly known as Automan).
People
Ringtail is a collaboration among computer scientists at The University of Michigan and Stanford University.
We collaborate closely with some great economists at the University of Michigan:
- Matthew D. Shapiro, Professor, Economics Department and Research Professor,
Institute for Social Research
- Margaret C. Levenstein, Research Scientist, Institute for Social Research and Adjunct Professor, Ross School of Business
- Mark Fontana, Ph.D. Candidate, Department of Economics
Technical Challenge
Building the Ringtail system has involved systems, user interaction, and statistical challenges. One of our current technical problems concerns the incredible difference in data scale between standard economics datasets and social media ones. In short, high-quality macroeconomic datasets are usually very small (at most, a data point each week), while low-quality social media data are very large (billions of data points a day). This massive difference in scale causes problems for many standard statistical techniques, such as feature selection. You can read more about this feature selection problem
in this WebDB paper.
Research Papers
Computer Science:
- Ringtail: Feature Selection for Easier Nowcasting. D Antenucci, MJ Cafarella, MC Levenstein, C Ré, M Shapiro. WebDB, 2013
- Ringtail: A Generalized Nowcasting System. D Antenucci, E Li, S Liu, B Zhang, MJ Cafarella, C Ré. Proceedings of the VLDB Endowment, 2013
Economics:
Last updated: March 16, 2014