Back to Home

FAQ & Troubleshooting

Find answers to common questions and solutions to common problems

General Questions
What is this platform for? ▼
The Energy Simulations Platform is a comprehensive tool for:
  • Energy price prediction - Forecast electricity prices using machine learning
  • Solar production estimation - Predict solar panel output based on location and weather
  • Battery optimization - Simulate optimal charging/discharging strategies
  • Risk analysis - Use Monte Carlo simulations to understand uncertainty
It's designed for energy analysts, homeowners with solar panels, and anyone interested in energy market optimization.
Is this platform free to use? ▼
Yes! The platform is completely free for personal and educational use. For commercial applications, please contact us for licensing options.
What languages are supported? ▼
The platform currently supports:
  • 🇬🇧 English (EN)
  • 🇩🇪 German (DE)
  • 🇪🇸 Spanish (ES)
You can switch languages using the globe icon in the top right corner of any page.
Data Questions
What file formats can I upload? ▼
Currently, we support CSV (Comma-Separated Values) files. The file should have:
  • A header row with column names
  • Comma or semicolon as delimiter
  • UTF-8 encoding (recommended)
Maximum file size is 50MB.
What columns are required in my dataset? ▼
It depends on what you want to predict:

For price prediction:
  • timestamp - Date/time column (ISO format preferred)
  • price - The price value to predict
  • Any additional numeric features (demand, renewable share, etc.)
For production prediction:
  • timestamp - Date/time column
  • production_kwh or similar - Production values
  • Weather data if available (temperature, irradiance, etc.)
Is my data secure? Where is it stored? ▼
Your data security is important to us:
  • Data is stored locally on the server you're running
  • Datasets marked as "private" are only accessible to you
  • You can delete your data at any time
  • No data is shared with third parties
If you're running the platform locally, all data stays on your machine.
Model Training
Which algorithm should I choose? ▼
Random Forest is recommended for most users:
  • ✓ Fast training time
  • ✓ Robust to outliers
  • ✓ Handles missing values well
  • ✓ Good accuracy out-of-the-box
Neural Network may be better if:
  • You have a very large dataset (10,000+ rows)
  • Complex non-linear patterns exist
  • You're willing to tune hyperparameters
How long does training take? ▼
Training time depends on:
  • Dataset size: 1,000 rows = ~30 seconds, 100,000 rows = ~5 minutes
  • Algorithm: Random Forest is faster than Neural Networks
  • Hyperparameters: More trees/layers = longer training
  • Hardware: Faster CPU/GPU = faster training
You can track progress in the Training page while it runs.
What do the training metrics mean? ▼
  • R² (R-squared): How much variance is explained. 1.0 = perfect, > 0.8 = good
  • RMSE: Root Mean Squared Error. Lower is better. Units match your target.
  • MAE: Mean Absolute Error. Average prediction error. Lower is better.

A model is generally considered "good" if R² > 0.7 and the RMSE/MAE are acceptable for your use case.
Predictions
Why are my predictions inaccurate? ▼
Common reasons for poor predictions:
  • Insufficient data: More training data usually = better predictions
  • Missing features: Important variables might not be in your dataset
  • Data drift: Training data is different from prediction data
  • Overfitting: Model memorized training data instead of learning patterns
Try training with more data or different features.
Can I make batch predictions via API? ▼
Yes! Use the /predict/batch endpoint:

POST /predict/batch with a JSON body containing an array of feature objects.

See the API Documentation for detailed examples.
Simulations
How does battery simulation work? ▼
The battery simulation models a virtual battery storage system:
  • Uses historical or predicted price data
  • Applies your specified strategy (e.g., buy low, sell high)
  • Considers battery capacity, efficiency, and degradation
  • Calculates total profit/savings over the simulation period
What is Monte Carlo simulation? ▼
Monte Carlo simulation runs your scenario hundreds of times with random variations to:
  • Understand the range of possible outcomes
  • Calculate probability of achieving target profits
  • Identify worst-case and best-case scenarios
  • Measure risk (Value at Risk, expected shortfall)
It's especially useful for understanding uncertainty in your energy strategy.
Troubleshooting

Upload fails with "Invalid file format"

The file couldn't be parsed as a valid CSV file.

Solution:

  1. Ensure the file has a .csv extension
  2. Open in a text editor and check the delimiter (comma or semicolon)
  3. Remove any BOM (byte order mark) if present
  4. Save as UTF-8 encoding
  5. Try uploading again

Training is very slow

Training is taking longer than expected.

Solution:

  1. Try using Random Forest instead of Neural Network
  2. Reduce the number of trees/layers
  3. Use a smaller subset of your data for initial testing
  4. Close other resource-intensive applications

Model not found error

Error: "Model file not found at specified path"

Solution:

  1. Ensure the model was trained successfully
  2. Check if you selected the correct model from the dropdown
  3. Verify the model file exists in models/own/ or models/public/
  4. Try refreshing the page and selecting the model again

Features mismatch error

"Input features don't match training features"

Solution:

  1. Check what features the model was trained with
  2. Ensure your input data has all required columns
  3. Column names must match exactly (case-sensitive)
  4. Order of columns doesn't matter

Solar map not loading

The solar potential map is blank or slow to load.

Solution:

  1. Check your internet connection (map tiles require network)
  2. Try refreshing the page
  3. Allow location access if prompted
  4. Try a different browser if issue persists

Still Need Help?

Can't find what you're looking for? Check out these resources:

Getting Started Guide API Documentation Swagger Docs