Home Pricing Archive About Privacy

A Data-Driven Framework for Predicting Solar Power Output Using Machine Learning and Time-Series Models

📑 5 slides 👁 2 views 📅 3/5/2026
0.0 (0 ratings)

Proposed Thesis Title

Solar PV energy is rapidly growing in global electricity systems.

Proposed Thesis Title
2

Research Problem & Questions

  • Solar forecasting is challenging due to weather variability.
  • Inaccurate forecasts lead to inefficient energy planning.
  • Key questions: ML accuracy, best short-term model, influential factors.
Research Problem & Questions
3

Hypothesis & Expected Direction

  • ML + time-series models outperform traditional statistical methods.
  • Random Forest & XGBoost capture nonlinear solar generation patterns.
  • Facebook Prophet effectively models seasonal and daily trends.
Hypothesis & Expected Direction
4

Data Sources

  • Public datasets: ENTSO-E Transparency Platform & OPSD.
  • Key variables: solar generation, wind, load, prices, time.
  • High-frequency European electricity data for ML forecasting.
Data Sources
5

Proposed Methodology

  • Approach: data preprocessing, feature engineering, model training.
  • Compare Random Forest, XGBoost, Prophet using RMSE/MAE/R².
  • Demonstrate continuous prediction with trained models.
Proposed Methodology
1 / 5