engleski

Temporal Analysis of Political Instability Through Descriptive Subgroup Discovery

The paper analyzes the Political Instability Task Force (PITF) dataset using a new methodology based on machine learning tools for subgroup discovery. While the PITF used static data, this study employs both static and dynamic descriptors covering the five-year period before onset. The methodology provides several descriptive models of countries especially prone to political instability. For the most part, these models corroborate the PITF’ s findings and support earlier theoretical works. The paper also shows the value of subgroup discovery as a tool for developing a unified concept of political instability as well as for similar research designs.

dynamic variables; political instability task force; subgroup discovery

Dealing with failed states: Crossing analytic boundaries. edited by Starr, H. Routledge London and New York, 2009, pp.94-107.