Parus Analytics

March-2024: Parus Analytics is going out of business and closing its doors forever!!

Yes, after 30 odd years of unrelenting snark, nearest-neighbor-based predictive models that never seemed to work, automated event coding systems that generally did, and numerous collaborations with a similar track record, Parus Analytics is folding its proverbail tent and shutting down! This web site remains for matters of historical record but, as the sayings go, so long and thanks for the fish, and...

Welcome

Software Development. Natural Language Processing. Statistical Analysis
Parus Analytics is a small independent consulting firm specializing in the development of software for the analysis of political events. We have done work on contracts for the U.S. National Science Foundation, the Defense Advanced Research Projects Agency, the U.S. multi-agency Political Instability Task Force (PITF) and the IARPA-sponsored Good Judgment Project.

Event Data

TABARI. CAMEO. PETRARCH. PHOENIX.
In early work at the University of Kansas and Pennsylvania State University, the research team at Parus pioneered the development of fully automated methods for the coding of political event data. These efforts included KEDS, the first successful automated coder, the CAMEO event and actor coding ontology, and TABARI, the coding program used in the research phases of the DARPA Integrated Conflict Early Warning System (ICEWS). Our recent efforts include providing support for the real-time PHOENIX data set, and development of the new PETRARCH coder, written in Python, hosted on GitHub as an open-collaboration project, and utilizing the Stanford CoreNLP natural language processing suite. In addition to the computational event data efforts, we also produce the PITF Worldwide Atrocities Data Set using machine-assisted coding methods.

eventdata.parusanalytics.com
This is the successor site for the KEDS project: follow these links to access presentations, papers, software and data sets.

Political Forecasting

Political Instability Task Force. Integrated Conflict Early Warning System.
Most of our efforts over the past twenty years have focused on statistical and computational methodologies for forecasting violent political conflict. Both PITF and ICEWS have demonstrated that systematic quantitative models, can consistently forecast a variety of indicators of political instability with accuracy in excess of 80% at policy-relevant forecast windows of six to twenty-four months. This is dramatically better than human forecasters—many of whom scarcely exceed the 50% accuracy of the proverbial "dart-throwing chimp"—and can be done using open-source data and methods.