Download Data
Electricity market data from ENTSOE, weather forecasts from Open-Meteo, and calendar features
Introduction
epftoolbox2
A Python library for electricity price forecasting with modular data pipelines and model evaluation. Epftoolbox2 is a continuation of the original epftoolbox project. It provides a modernized, modular architecture while maintaining the core focus on electricity price forecasting research and applications.
What is epftoolbox2?
Process Data
Resample, create lag features, convert timezones with flexible transformers
Validate Quality
Check for nulls, continuity gaps, and view EDA statistics
Train Models
OLS and LassoCV models with rolling-window evaluation
Multi-Horizon Forecasting
Forecast multiple days ahead (1-7+ days) with horizon-specific evaluation
Export Reports
Excel with conditional formatting or rich terminal tables
Architecture
The library follows a modular design where each component (sources, transformers, validators, models, evaluators, exporters) can be used independently or combined into automated pipelines that handle everything from data fetching to report exporting.
Key Features
Fluent API
Method chaining for clean, readable pipeline construction
Built-in Caching
Avoid redundant API calls with automatic data caching
Incremental Training
Resume interrupted model training runs
GIL-Free Multithreading
True parallel execution with Python 3.13t+
Extensible
Easy to add custom sources, transformers, models, etc.
AI Ready
Documentation for AI agents via llms.txt
Next Steps
Installation Install the library and configure your environment
Quick Start Run your first electricity price forecast