A web application for PV energy forecast using meteorological data.
This web application provides a PV (Photovoltaic) energy forecast based on meteorological data. It leverages machine learning algorithms to predict solar energy production for a given location and time frame. The forecast can help users optimize energy consumption and make informed decisions.
- Accurate PV energy forecast based on meteorological data
- User-friendly interface with intuitive controls
- Customizable parameters for EDA
- Historical data analysis for performance evaluation
- Real-time hourly forecast for the next 3 days using a weather API
The web app can be accessed at the following link: https://pv-energy-forecast.streamlit.app/
- Home: this page features a PV panel location explorer and an interactive dataset explorer.
- EDA: Exploratory Data Analysis with an interactive way to choose the data granularity and daytime (07:00 AM - 18:00 PM) or full time dataset.
- ML Model Estimation: this page features metrics for the three ML models (MLP, XGBoost and Random Forest), deviation plots and feature value importance using SHAP VALUES
- Forecast: in this page the MLP model forecast can be viewed for the next 3 days; the forecast is retrieved from the weather API.
Credits to https://dkasolarcentre.com.au/ for providing the PV energy historical dataset.