Development of an Automated Conflict Prediction System. State Space ARIMA Methods

Inhaltsverzeichnis (Table of Contents)

• Abstract
• Lists of Abbreviations and Symbols
• 1 Introduction
• 2 Method
  • 2.1 State Space Modelling Approach
    • 2.1.1 Linear Gaussian State Space Model
    • 2.1.2 ARIMA State Space Model
    • 2.1.3 SARIMA State Space Model
    • 2.1.4 (S) ARIMAX Regression with (S) ARIMA errors
    • 2.1.5 Kalman Filter
  • 2.2 No-Change Baseline Model
  • 2.3 Evaluation Metrics
    • 2.3.1 TADDA
    • 2.3.2 Mean Absolute Error (MAE)
    • 2.3.3 Root Mean Square Error (RMSE)
• 3 Data
  • 3.1 Armed Conflict Location and Event Data Project (ACLED)
    • 3.1.1 Analysis of Conflict Incidence and Country Categorization
  • 3.2 International Monetary Fund - World Economic Outlook Database (IMF-WEO)
  • 3.3 World Bank World Development Indicators (WB-WDI)
  • 3.4 Variable Overview and Missing Data
• 4 Implementation in Python
  • 4.1 No-Change Forecaster Class
  • 4.2 State Space ARIMA Forecaster Class
  • 4.3 Grid Search Class
  • 4.4 Automated Model Building Process
• 5 Results
  • 5.1 Global Model Performances
  • 5.2 Country-Level Model Performances

Zielsetzung und Themenschwerpunkte (Objectives and Key Themes)

This thesis aims to develop a fully automated machine learning framework for optimizing econometric state space ARIMA models in a data-driven manner. The framework compares predictions of a model portfolio, including standard ARIMA, seasonal SARIMA, ARIMAX models with socio-economic variables, and ARIMAX models with conflict indicators. The work addresses the challenge of predicting conflict onset events, which are rare and difficult to detect using traditional machine learning techniques.

• Development of an automated machine learning framework for optimizing state space ARIMA models
• Comparison of different ARIMA model variations for conflict prediction
• Evaluation of model performance using out-of-sample prediction errors
• Assessment of the ability to outperform naïve heuristics in conflict prediction
• Analysis of country-level model performances

Zusammenfassung der Kapitel (Chapter Summaries)

The Introduction chapter sets the context for the research by discussing the urgency for early conflict detection and the challenges of existing prediction systems. Chapter 2 delves into the methodology, explaining the state space modeling approach, the specific ARIMA models used, the evaluation metrics, and the no-change baseline model. Chapter 3 focuses on the data used, describing the sources, selection process, and variable overview. Chapter 4 details the implementation of the machine learning framework in Python, including the different forecaster classes and the automated model building process. Finally, Chapter 5 presents the results, comparing the performance of the different models and analyzing their predictive capabilities at both the global and country levels.

Schlüsselwörter (Keywords)

This thesis focuses on the following keywords: conflict prediction, state space ARIMA models, automated machine learning, model portfolio, out-of-sample prediction errors, naïve heuristics, conflict incidence, socio-economic indicators, conflict indicators, TADDA score, ACLED data, IMF World Economic Outlook Database, World Bank World Development Indicators.