Fausty099/SILO-oil-community-risk

AI-powered early warning system predicting community-company conflict escalation in oil-producing regions of West Africa using satellite flare data, conflict signals and machine learning. # SILO ### Social Impact & License to Operate Intelligence System *An AI-powered early warning system predicting community-company conflict escalation* *in oil-producing regions of West Africa* --- ## The Problem Nobody Has Solved Nigeria's Niger Delta produces over 90% of the country's oil revenue. It is also one of the most conflict-affected regions in sub-Saharan Africa. When oil companies lose the trust of host communities — what researchers call the **Social License to Operate (SLO)** — the consequences escalate fast: protests turn to pipeline bombings, production halts, and billions in deferred output. The 2016 Niger Delta Avengers campaign alone dropped Nigeria's oil output to a **30-year low**. The warning signs were there years before. Gas flaring rising. Protests accelerating. Fatalities spiking. **No system existed to connect those dots.** SILO is that system. --- ## What SILO Does SILO fuses three publicly available data streams into a single machine learning pipeline that predicts **which state will enter conflict escalation within the next year** — before it happens. ``` NASA VIIRS Satellite Data → gas flare intensity & volume Community Conflict Signals → protests, riots, violence, battles Engineered Trend Features → lag effects, anomalies, acceleration ↓ XGBoost Classifier ↓ SLO Erosion Risk Score (0–100%) + SHAP Explanation + Interactive Risk Map ``` **Output example:** > *"Edo State has a 97.8% probability of entering conflict escalation — driven by rapidly accelerating conflict (+0.94 SHAP) and sustained protest activity (+0.62 SHAP)."* That's not just a number. It's a reason. That's what makes SILO actionable. --- ## Why This Is Novel | What Already Exists | SILO | |---|---| | Oil spill detection via satellite | Widely studied | | Community health impacts (qualitative) | Widely studied | | General political conflict prediction | Widely studied | | Resource curse economics (macro-level) | Widely studied | …