25-ID2223-mlfs-book
Finland Energy Consumption Dashboard
Finland Energy Consumption Forecast
Real-time energy consumption forecasting powered by Fingrid API and machine learning
7-Day Forecast
Model Performance on Test Data
Hindcast - Predictions
Compares 1-day-ahead predictions from baseline model against actual consumption values over historical periods. This validates model’s ability to predict consumption patterns.
Model Analysis - Baseline Model
Feature Importance
Feature Correlation
Model Analysis - Lag Models
Feature Importance Comparison
Feature Correlation Comparison
This dashboard displays the 7-day forecast for Finland’s electricity consumption using:
- Fingrid API: Real-time energy consumption data (3-minute intervals)
- Open-Meteo API: Weather forecast data (temperature, wind, precipitation, etc.)
- XGBoost Model: Machine learning model trained on historical consumption patterns
The forecast is updated daily at 6:15 AM UTC through automated GitHub Actions workflows.
Data Sources
- Energy Data: Fingrid Open Data Hub - Dataset 193 (Electricity consumption in Finland)
- Weather Data: Open-Meteo Forecast API (Helsinki coordinates)
- Feature Store: Hopsworks Feature Store for data versioning and orchestration
Model Features
The model uses temporal and weather features:
- Temporal: hour, day, month, day of week, weekend indicator, week of year (with cyclic encoding)
- Weather: temperature, wind speed/direction, precipitation, cloud cover, pressure, solar radiation