geodesmond1990-design/Machine-Learning-Assessment-of-Oil-Spill-Containment-Efficiency-and-Response-Time-in-the-Niger-Delta

An open-source Realtime video/audio conferencing web app with features like chat, meeting notes, summary, sharing files... # Niger Delta Oil Spill - Containment Efficiency and Response Time ML Analysis -green) --- **Title:** Quantifying Spill Containment Efficiency and Response Time Disparities in Niger Delta Oil Infrastructure Using Machine Learning: A Comparative Analysis of Operators and Facility Types **Status:** Under Review --- ## Overview This repository contains the complete, reproducible analysis for Paper 3, which presents the **first ML-based quantitative analysis of spill containment efficiency and response time** for the Niger Delta — defining CER and RTI as explicit operational performance metrics and identifying the predictors of poor containment outcomes. **Core innovation:** We transform mandatory NOSDRA incident reports into actionable operational performance metrics (CER and RTI), demonstrate that flowline facilities have a statistically significant containment efficiency deficit versus pipelines, and identify a structural response time plateau since 2020 using Chow structural break testing. ### Key Results **Containment Efficiency Ratio (CER):** - Mean CER = 41.2% (SD 38.4%), Median = 18.5% - Most incidents achieve low-to-moderate recovery — this is the norm, not an exception **Flowline Deficit (most actionable finding):** - Pipeline CER = 45.1% - Flowline CER = **28.3%** - Mann-Whitney p = **0.002**, Cohen d = **0.41** (small-medium effect) - This difference is independent of volume and surface type (confirmed by multivariate regression) **Operator Comparison:** - NAOC CER = 42.3% vs SPDC CER = 38.7% - Mann-Whitney p = 0.14 — **no statistically significant difference** after controlling for confounders **Response Time (RTI):** - RTI improved **18%** from 2016 to 2020 (from ~11.2 days to ~6.8 days) - Chow test confirms structural break at Q2 2020 (F=6.84, **p=0.009**) - RTI has **plateaued** since 2020 — structural constraint identified **ML Performance (CER classification, 5-fold CV):** | Model | Accuracy | F1 | |-------|----------|----| | KNN …