🌍 Global Terrorism Analysis: Big Data Insights Using PySpark and Tableau

📊 Overview

This project explores global terrorism trends using big data analytics and visualization tools. Leveraging the Global Terrorism Database (GTD) — a comprehensive dataset of terrorist incidents from 1970 to 2017 — this analysis aims to uncover patterns, hotspots, and predictive insights to aid global counterterrorism strategies.

🎯 Objectives

• Analyze over 180,000 terrorism records using PySpark for scalable data processing.
• Use Linear Regression to predict casualties based on attack and weapon types.
• Apply K-Means Clustering to detect geographic terrorism hotspots.
• Visualize insights using Tableau and Python libraries like Matplotlib, Seaborn, and Folium.

📂 Project Structure

📁 Global_Terrorism_Analysis/
├── 📄 Big_Data_Analysis_and_Data_Visualization.ipynb  # Jupyter Notebook with code and visualizations
├── 📄 Global_Terrorism_Analysis.docx                  # Full research paper/report
├── 📄 terrorism_map.html                              # Interactive map of attacks
├── 📄 choropleth_attacks_by_country.html              # Choropleth map of attacks by country
├── 📄 choropleth_fatalities_by_country.html           # Choropleth map of fatalities
├── 📄 heatmap_attacks.html                            # Heatmap of attacks
├── 📄 map_by_target_type.html                         # Map by target type
├── 📄 terrorism_dashboard.twbx                         # Tableau dashboard (if applicable)

🔧 Tools & Technologies

Tool	Purpose
PySpark	Distributed data processing and ML modeling
Python	Data wrangling and visualizations
Tableau	Interactive and dynamic data dashboards
Folium	Geospatial mapping of attacks
Seaborn & Matplotlib	Statistical and trend plotting

🧠 Key Analyses

📈 Linear Regression

• Goal: Predict number of casualties using attacktype1 and weaptype1.
• Features engineered: StringIndexing, VectorAssembler
• Metrics: R² Score and RMSE

📍 K-Means Clustering

• Goal: Group attacks based on latitude & longitude to identify terrorism hotspots.
• Clusters: Set to 5 for regional analysis.
• Normalization: StandardScaler used for geographic feature scaling.

🗺️ Visualizations

• Temporal trends of terrorist activity over time.
• Attack type and region-based casualty summaries.
• Word clouds, choropleths, and interactive maps to identify and analyze high-risk areas.

📊 Tableau Dashboard

A dynamic Tableau dashboard provides:

• Yearly trends in terrorism
• Interactive geographical maps of attacks
• Top terrorist groups
• Regional casualty breakdowns
• Attack methods and frequencies

📌 Figures available in the report and attached .twbx file (if included).

📁 Dataset

• Source: Global Terrorism Database on Kaggle
• Records: 180,000+ terrorist events from 1970–2017
• Features: 135 attributes including location, group, attack type, weapon, target, casualties, etc.

📌 How to Run

1. Install dependencies:

   pip install pyspark pandas seaborn matplotlib folium wordcloud

2. Run the Jupyter Notebook:

   jupyter notebook Big_Data_Analysis_and_Data_Visualization.ipynb

3. View .html maps in your browser.

4. Open Tableau .twbx dashboard for dynamic exploration.

🔍 Insights & Conclusions

• Bombings and armed assaults are the most common attack types.
• Middle East, South Asia, and parts of Africa emerge as key hotspots.
• While regression models highlight useful predictors, additional variables like target type and socio-political factors could enhance accuracy.

💡 Future Work

• Incorporate temporal clustering (by time of day, month).
• Add socio-economic, political, or organizational affiliation features.
• Explore deeper networks within terrorist organizations.

🤝 Social Impact

By analyzing terrorism trends using data, this project aims to:

• Assist governments and security agencies in proactive counterterrorism.
• Provide humanitarian insights for resource allocation.
• Demonstrate the power of data science in addressing global security challenges.

🧾 License

This project is for educational and research purposes. Please cite appropriately if referenced.