BKO AI and Texas A&M developed a model to accurately predict next-day wind energy availability, significantly enhancing market participation and decision-making.
A wind energy provider needed a more accurate prediction of next-day energy availability when participating in “next-day market” activity. The ability to accurately predict next-day energy availability will allow decision makers to plan ahead. The more accurate the models the more informed the decision and profitability.
In partnership with Texas A&M, BKO AI implemented pooled, un-pooled, and hierarchical models using Bayesian methods. Utilized Python, Minitab, R and JMP to clean and analyze data to formulate an accurate model for predicting the next-day energy available to sell on the market.
Produced more accurate prediction results with Partial-Pooling Bayesian hierarchical model. Additional accuracy created additional value of $27,049 per 5 minutes of operating time and approximately $38 million cost savings.
