Data-based Computational Approaches to Forecasting Political Violence

Schrodt, Philip A.; Yonamine, James E.; Bagozzi, Benjamin E.

doi:10.1007/978-1-4614-5311-6_7

Cited by 22 publications

(25 citation statements)

References 122 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The tree structure itself is critical to the analysis of comparing two data sets as we are doing in this article. Other methods, such as regression analysis (Schrodt, Yonamine, and Bagozzi, ), neural networks (Schmidhuber, ), random forests (Breiman, ), and similar prediction models may provide better predictions, but that is not the goal of this analysis. Regression analysis, neural networks, random forests, and others do not provide the characterization of the dependent variable values in terms of the independent variables and their values necessary for the comparison analyses demonstrated in this article.…”

Section: How Classification Trees Workmentioning

confidence: 99%

“…Schrodt, Yonamine, and Bagozzi (:145) identify four specific ways in which algorithmic approaches may be better suited than statistical models for analyzing terrorism. First, because algorithmic techniques are not sensitive to degrees of freedom, they are generally better at handling “big data” with large quantities of independent variables, including many conflict and terrorism databases used by social scientists.…”

Section: Using Classification Analysis To Study Terrorismmentioning

confidence: 99%

See 1 more Smart Citation

Comparing Classification Trees to Discern Patterns of Terrorism*

Saiya

Scime

2019

Social Science Quarterly

View full text Add to dashboard Cite

Objective. Though applied widely in the fields of medicine, finance, ecology, psychology, and computer science, machine learning algorithmic-based methods are a relatively novel approach to social scientific analysis that have yet to be extensively applied. Yet as we argue in this article, a specific form of algorithmic analysis known as C4.5 classification trees has much to offer social analysis and, specifically, the study of social and political violence. Method. This article describes four novel classification model comparison techniques for the C4.5 classification method and applies them to the study of terrorism. Results. Our state-level analysis suggests that there is something fundamentally different in the targeting choices of religious and secular terrorists. Conclusion. This analysis highlights the ability of classification trees to heighten our understanding of terrorism and even provide recommendations to policymakers for avoiding future attacks.

show abstract

Section: How Classification Trees Workmentioning

confidence: 99%

Section: Using Classification Analysis To Study Terrorismmentioning

confidence: 99%

Comparing Classification Trees to Discern Patterns of Terrorism*

Saiya

Scime

2019

Social Science Quarterly

View full text Add to dashboard Cite

show abstract

“…Figure 2 represents the top frequent subgraphs associated with DPCs in Brazil, Colombia, Mexico, and Venezuela. The thick edges represent more adversarial interactions (event types [11][12][13][14][15][16][17][18][19][20] whereas the thin edges represent cooperative interactions (event types 01-10). The figure shows that, in Colombia, conflicts between Rebellion, Military, and Government are frequent during domestic crises.…”

Section: Classifying Dpc Eventsmentioning

confidence: 99%

“…Methods used include discriminant analysis [17], HMMs [16], Bayesian time series forecasting [14,5,19,11], and vector auto regression (VAR) methods [10,6]. For a survey on these predictive models, we point the readers to [18].…”

Section: Introductionmentioning

confidence: 99%

Detecting and forecasting domestic political crises

Keneshloo

Cadena

Korkmaz

et al. 2014

Proceedings of the 2014 ACM Conference on Web Science

View full text Add to dashboard Cite

Forecasting a domestic political crisis (DPC) in a country of interest is a very useful tool for social scientists and policy makers. A wealth of event data is now available for historical as well as prospective analysis. Using the publicly available GDELT dataset, we illustrate the use of frequent subgraph mining to identify signatures preceding DPCs, and the predictive utility of these signatures through both qualitative and quantitative results.

show abstract

“…insights) from conflict management databases containing details of international conflicts. In a survey article, Schrodt et al [28], describe a range of techniques that have been used for conflict prediction from large datasets including treebased algorithms, clustering approaches using Latent Dirichlet Analysis and hidden markov models. This paper contributes to the analysis branch, where insights from near real-time data in GDELT are obtained from disparate news sources.…”

Section: Related Workmentioning

confidence: 99%

Multi-level Analysis of Peace and Conflict Data in GDELT

Keertipati

Savarimuthu

Purvis

et al. 2014

Proceedings of the MLSDA 2014 2nd Workshop on Machine Learning for Sensory Data Analysis

View full text Add to dashboard Cite

Huge amounts of news items are being published from various sources around the world. Our interest in this paper is from the perspective of peace keeping organizations such as United Nations and Red Cross that can benefit from information aggregated from different sources to make important decisions. GDELT project is one such source where data about conflicts is aggregated from various newspaper sources around the globe. This paper makes three contributions to the analysis of peace and conflict data available on GDELT. First, the paper presents two case studies to investigate whether the data extracted from news items in fact captures the global events accurately. The case studies considered were the Sri Lankan civil war, and the 2006 Fijian coup. Second, in these case studies, the study demonstrates how change point analysis can be used to identify important trends which will be valuable for making decisions. Third, based on experience gained through the exploration of these two case studies, it proposes a conceptual model of a multi-level analysis in the form of a dashboard system that would be of interest to different stakeholders.

show abstract

Data-based Computational Approaches to Forecasting Political Violence

Cited by 22 publications

References 122 publications

Comparing Classification Trees to Discern Patterns of Terrorism*

Comparing Classification Trees to Discern Patterns of Terrorism*

Detecting and forecasting domestic political crises

Multi-level Analysis of Peace and Conflict Data in GDELT

Contact Info

Product

Resources

About