How youssef595/hourly-electricity-price-forecasting works
Project type and purpose
• This is a data-science/time-series forecasting project (“Hourly-electricity-price-forecasting”) aimed at predicting hourly electricity prices in Spain using historical energy and weather data. It uses various modeling strategies (base boosting models, multiple train/test splits, walk-forward validation, hybrid multivariate forecasting) to produce reports and presentations.
File structure
• Top-level directories:
• Data: raw CSV datasets (energy_dataset.csv, weather_features.csv)
• Notebooks: Jupyter notebooks for exploration, feature engineering, modeling, and visualization
• Presentation: final PowerPoint
• Reports: PDF/HTML reports and model performance summaries
• README.md describes data sources (ENTSOE, Red Eléctrica España), approaches, and visualizations.
• There are no separate “src” or “backend” directories—computation is entirely in notebooks.
README for additional insights
• Datasets: two main CSVs (energy & weather), concatenated for modeling
• Approaches:
a. Base implementation (2015–2017 training, 2018 testing)
b. Multiple train/test splits
c. Walk-forward validation
d. Hybrid multivariate forecasting (varying time lags)