Building a prototype predictive model to forecast solar panel electricity generation

To participate in the electricity exchange trading,  or customer needs an exact forecast of solar panel electricity generation, aiming to reduce financial losses.

We performed a fundamental analysis of the client’s data and enriched it with additional weather and relevant data from various sources to improve predictive model accuracy. We built several model prototypes using Cloudera Data Science Workbench and various machine learning algorithms such as decision trees, neural networks, and metric-based algorithms.

The solution significantly improved the accuracy of solar panel electricity generation forecasts (over 95%), reducing the potential financial losses associated with forecast errors in exchange trading.