Identifying the complex causes of civil war : a machine learning approach / / Atin Basuchoudhary [and three others]

• Autore: Basuchoudhary Atin
• Pubblicazione: Cham, Switzerland : , : Palgrave Pivot, , [2021]; ©2021
• Descrizione fisica: 1 online resource (137 pages)
• Disciplina: 303.64
• Soggetto topico: Civil war - Forecasting; Civil war - Data processing; Civil war - Prevention
• Nota di contenuto: Intro -- Contents -- List of Figures -- List of Tables -- 1 Prologue: Why This Book? -- References -- 2 Data Description and Preliminary Processing -- 2.1 Data Sources -- 2.2 Data Preparation -- 2.3 Data Imputation -- 2.4 Recursive Feature Elimination with Random Forests -- References -- 3 Methodological Workflow -- 3.1 The Idea -- 3.2 Empirically Informed Covariate Selection (EICS) -- 3.2.1 Identifying Risk and Confounder Variables -- 3.2.2 Eliminating Colliders and Intermediate Variables -- 3.3 Causal Results -- 3.3.1 Partial Dependence Plots -- 3.3.2 The Most Accurate Partial Dependence Plot -- 3.3.3 The Complete Methodological Workflow -- 3.4 Takeaways -- References -- 4 Constitutional Changes and Civil War -- 4.1 Empirically Informed Covariate Selection Results -- 4.2 Do Constitutional Changes Cause Civil War? -- 4.3 Data-Driven Theory. How Might Constitutional Changes Cause a Civil War? -- 4.4 Takeaways -- References -- 5 Infant Mortality, State Capacity, Rents, and Civil War -- 5.1 EICS Results -- 5.2 Causal Analysis -- 5.3 Takeaways -- References -- 6 Foreign Aid and Civil Conflict -- 6.1 EICS and Specifying a Causal Model -- 6.2 Net Secondary Income Causes Civil Conflict? -- 6.3 Takeaways -- References -- 7 Demonstrations, Grievance, and Civil Conflict -- 7.1 EICS, Model Specification, Anti-government Demonstrations -- 7.2 Peaceful Demonstrations Cause Civil Conflict -- 7.3 Takeaways -- References -- 8 Epilogue -- 8.1 Insights -- 8.1.1 Methodological Insights: -- 8.1.2 Implications for the Civil Conflict Literature -- Appendices -- Appendix 1 -- The 261 Variables Considered for Feature Extraction -- Appendix 2: The Code -- A2.1 Pre-processing the Data -- Random Forest Missing Value Imputation -- Creating Five-Year Moving Averages and Non-overlapping Lags -- Splitting into Learning and Test Samples Stratified by Conflict History. ; Recursive Feature Elimination -- A2.2. Detecting Causal Pathways: Empirically-Informed Covariate Selection -- Causal Graph Function -- Detecting the Causal Pathways of the Top-30 Variables -- A2.3. Causal Analysis Models -- Setup -- Logistic Regression with Controls -- Inverse Probability of Treatment Weights Balancing -- Bayesian Additive Regression Trees -- Bayesian Additive Regression Trees with IPTW Propensity Score -- References -- Index.
• Titolo autorizzato: Identifying the complex causes of civil war
• ISBN: 3-030-81993-0
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 9910506391803321
• Lo trovi qui: Univ. Federico II