Predicting the Unpredictable: Reproducible BiLSTM Forecasting of Incident Counts in the Global Terrorism Database (GTD)
Oluwasegun AdegokeAbstract
A Bidirectional LSTM outperforms classical and deep learning baselines in forecasting weekly terrorism incidents using the Global Terrorism Database.
We study short-horizon forecasting of weekly terrorism incident counts using the Global Terrorism Database (GTD, 1970--2016). We build a reproducible pipeline with fixed time-based splits and evaluate a Bidirectional LSTM (BiLSTM) against strong classical anchors (seasonal-naive, linear/ARIMA) and a deep LSTM-Attention baseline. On the held-out test set, the BiLSTM attains RMSE 6.38, outperforming LSTM-Attention (9.19; +30.6\%) and a linear lag-regression baseline (+35.4\% RMSE gain), with parallel improvements in MAE and MAPE. Ablations varying temporal memory, training-history length, spatial grain, lookback size, and feature groups show that models trained on long historical data generalize best; a moderate lookback (20--30 weeks) provides strong context; and bidirectional encoding is critical for capturing both build-up and aftermath patterns within the window. Feature-group analysis indicates that short-horizon structure (lagged counts and rolling statistics) contributes most, with geographic and casualty features adding incremental lift. We release code, configs, and compact result tables, and provide a data/ethics statement documenting GTD licensing and research-only use. Overall, the study offers a transparent, baseline-beating reference for GTD incident forecasting.
Community
🚨 New: Bidirectional LSTM for Terrorism Forecasting (37% improvement over baselines)
Demonstrate that BiLSTM networks significantly outperform classical forecasting methods on the Global Terrorism Database (1970-2016), achieving RMSE 6.19 vs 9.89 for linear regression.
🔬 Key Contributions:
• Comprehensive ablation studies (15+ configurations tested)
• Systematic analysis of historical data requirements (5yr → 46yr)
• Feature importance quantification (rolling statistics are critical: +66% impact)
• Optimal sequence length identification (30-week lookback)
📊 Main Finding: Bidirectional processing captures temporal dependencies that unidirectional LSTMs miss entirely. LSTM+Attention (RMSE 9.19) performed worse than simple BiLSTM (6.19), highlighting that architecture choice matters more than complexity.
🔓 Fully Reproducible:
✅ Complete code & processed data on Zenodo (DOI: 10.5281/zenodo.17409540)
✅ All ablation experiments reproducible (~2 hrs on standard laptop)
✅ Trained model weights included
💡 Why This Matters: First systematic deep learning study on GTD with rigorous ablations, demonstrating that 20+ years of historical data is critical for reliable forecasting (5yr data → 220% performance collapse).
🔗 Code: github.com/Davidavid45/Deep-Learning-in-Counterterrorism
🔗 Zenodo: https://doi.org/10.5281/zenodo.17409540
#DeepLearning #TimeSeries #LSTM #ReproducibleResearch #SecurityAnalytics
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- Attention-Enhanced LSTM Modeling for Improved Temperature and Rainfall Forecasting in Bangladesh (2025)
- Forecasting Russian Equipment Losses Using Time Series and Deep Learning Models (2025)
- A Comprehensive Forecasting-Based Framework for Time Series Anomaly Detection: Benchmarking on the Numenta Anomaly Benchmark (NAB) (2025)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment:
@librarian-bot
recommend
Models citing this paper 1
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper