Tech talk: Building actionable demand forecasting systems

From fast food chains trying to avoid a chicken nugget shortage, to the NHS trying to stay one step ahead of a pandemic, good demand forecasting is what keeps the world running smoothly. 

Machine learning and statistics can be powerful systems for anticipating demand, but they’re not infallible. It’s no mean feat to extract demand insights that are accurate, detailed enough to be actionable, and easy for non-technical teams to understand.

In this talk, Faculty’s Director of AI, Ilya Feige, and R&D Lead Tom Begley will demonstrate how we build demand forecasting tools that organisations can use and trust. We’ll cover: 

• How well-calibrated uncertainty estimates lead to more actionable insights.
• How explainability helps end users trust and act on AI-generated forecasts.
• How exploiting the hierarchical structure of your data can solve the challenges created by small sample sizes. 

You’ll need a basic understanding of machine learning and statistics to get the most value from this talk.