Tosa-omokhoa/SPE-Conference-Lagos-NAICE2026-1017

Companion code, dataset and figures for SPE-NAICE2026-1017 — A Bradley Curve-AI Framework for Systematic Safety Culture Improvement in Nigeria's Upstream Oil and Gas Industry # A Bradley Curve-AI Framework for Systematic Safety Culture Improvement in Nigeria's Upstream Oil and Gas Industry **Paper ID:** SPE-NAICE2026-1017 **Conference:** SPE Nigeria Annual International Conference and Exhibition (NAICE 2026), Lagos, Nigeria, 3–5 August 2026 **Author:** Omokhoa Oshose Tosayoname, University of Nigeria, Nsukka --- ## Overview This repository contains all computational materials supporting the above paper — the dataset, the machine learning models, the figures, and the companion Google Colaboratory notebook referenced as Appendix A in the manuscript. The paper presents a four-phase integrated framework that combines the DuPont Bradley Curve safety culture maturity model with three machine learning algorithms to create a data-driven pathway for systematic HSE improvement in Nigerian upstream oil and gas operations. --- ## Key Findings - **Bradley Curve stage accounts for 52.9% of total LTIF predictive power** (SHAP analysis of XGBoost model) — more than all other operational and technical features combined. This is the first quantitative demonstration of this finding using Nigerian upstream data. - **51% of Nigerian upstream operational records** sit at the Reactive or Dependent stage (the two lowest cultural maturity levels). - A **24.7% LTIF reduction** is achievable within one operational cycle by improving three leading indicators that require no capital investment: PTW compliance, leadership safety observation frequency, and near-miss reporting rate. - An original **Environmental Incident Intensity Index (EIII)** is introduced, normalising spill volumes against production output to give a more honest environmental performance signal than raw volumetric reporting. --- ## Repository Structure ``` 📁 repo/ ├── README.md ← This file ├── notebook/ │ └── NAICE2026_SPE1017_Colab.ipynb ← Full Python implementation (Appendix A) ├── data/ │ └── master_hse_dataset.csv ← 264-record engineered dataset …