Improving Country Conflict and Peace Modeling: Datasets, Imputations, and Hierarchical Clustering
Date of Award
Doctor of Philosophy (PhD)
Department of Operational Sciences
Darryl K. Ahner, PhD
Many disparate datasets exist that provide country attributes covering political, economic, and social aspects. Unfortunately, this data often does not include all countries nor is the data complete for those countries included, as measured by the dataset’s missingness. This research addresses these dataset shortfalls in predicting country instability by considering country attributes in all aspects as well as in greater thresholds of missingness. First, a structured summary of past research is presented framed by a developed casual taxonomy and functional ontology. Additionally, a novel imputation technique for very large datasets is presented to account for moderate missingness in the expanded dataset. This method is further extended to establish the MASS-impute algorithm, a multicollinearity applied stepwise stochastic imputation method that overcomes numerical problems present in preferred commercial packages. Finally, the imputed datasets with 932 variables are used to develop a hierarchical clustering approach that accounts for geographic and cultural influences that are desired in the practical use of modeling country conflict. These additional insights and tools provide a basis for improving future country conflict and peace research.
DTIC Accession Number
Leiby, Benjamin D., "Improving Country Conflict and Peace Modeling: Datasets, Imputations, and Hierarchical Clustering" (2022). Theses and Dissertations. 5543.