Time Series Forecasting for Solar Energy Generation in Germany

The project predicts weekly solar energy generation in Germany for 2021-2022 based on 2015-2020 data with accuracy of 81% using seasonal ARIMA model.

Libraries: statsmodels, datetime, scikit-learn, matplotlib, seaborn, pandas, numpy

Table of contents

• Dataset
• Machine Learning problems

Dataset

• The dataset of renewable energy generation is from data.open-power-system-data.org (original source: ENTSO-E Transparency).

Machine Learning problems

• The problem is Time Series Forecasting for solar energy generation in Germany for 2021-2022 based on 2015-2020 data. The given problem was solved by using a Seasonal ARIMA model from statsmodels with accuracy of 84%. The result is in the DE_solar_generation.ipynb notebook.