Husayn01/SPE-DSEATS-Africa-Datathon-2026

# SPE DSEATS Africa Datathon 2026 Final submission workspace for the SPE DSEATS Africa Region 2026 oil-presence classification challenge. ## Project Structure - `notebooks/Prometheus_PythonCode.ipynb` - consolidated submission notebook for Part 1 and Part 2. - `predictions/Prometheus_Prediction_Part1.csv` - current standard ML prediction file. - `predictions/Prometheus_Prediction_Part2.csv` - current domain-informed prediction file. - `reports/Prometheus_Presentation.pptx` - current presentation export. - `data/raw/` - local copies of the original challenge CSV/PDF files. ## Data The final notebook checks local CSV files in `data/raw/` first, then tries Google Drive, then tries token-free GitHub raw URLs as backups. This makes the notebook runnable in Jupyter with local data or in Google Colab with the Drive links. The notebook preserves `Trap_Type = "None"` as a real no-trap category by loading CSVs with: ```python pd.read_csv(url, keep_default_na=False, na_values=[""]) ``` ## Running the Notebook Locally, install dependencies from `requirements.txt` and run the notebook in Jupyter. In Google Colab, open `notebooks/Prometheus_PythonCode.ipynb` and run from the top for the submission workflow. The notebook includes a package-install cell for optional modeling libraries and remote CSV loading for the two datasets. The current prediction CSVs each contain 2,000 binary predictions in a single `Oil_Presence` column.