Energy Consumption & Generation Prediction Based on Weather Conditions

As a team leader and team member, my part of the project focused on predicting renewable energy consumption and generation in Spain using machine learning. I analyzed the Kaggle dataset "Hourly Energy Demand Generation and Weather" with algorithms like Gradient Boosting Regressor and Random Forest Regressor. By isolating energy sources and using hourly data, we achieved high accuracy. The analysis showed that weather conditions and historical data are strong predictors of energy trends. This work highlights the potential of renewable energy as a greener alternative to fossil fuels.

​

Final Outcome:

The research demonstrated that weather conditions and historical data strongly predict energy generation and consumption. Nonrenewable energy generation and prices were accurately predicted from weather conditions. High accuracy in predicting renewable energy generation was achieved by isolating energy sources (e.g., solar energy) and aggregating data by month. Predicting total renewable energy generation based solely on historical data proved challenging due to numerous external factors affecting renewable energy sources.

Link for Code

Link for Report

- Python (Pandas, NumPy, Seaborn, Matplotlib, Scikit-Learn, Scipy)

- SQL

- Machine Learning (Gradient Boosting Regressor, Random Forest Regressor, Voting Regressor, Linear Regression, Ridge Regression, PCA)

- Techniques (Data Preprocessing, Feature Engineering, Outlier Detection, Correlation Analysis, Hyperparameter Tuning, Cross-Validation)

- Data Analysis (EDA, Visualizations)