Multi-horizon forecasting and associated challenges

Multi-horizon forecasting is essential to many forecasting use cases, but brings a range of challenges with it. Many techniques are promising when forecasting a single point, but suffer from quick divergence (recurrent techniques) and/or conflicting optimization problems (multi-output models) when extended to forecast a multi-point horizon.

Let’s look at these challenges, why they occur, and what TIM does differently to overcome these obstacles.

 

 

---

 

Time for Time Series dives into specific features and looks at general functionalities in TIM, as well as covering many topics related to time-series analysis. We welcome everyone, from managers to data scientists, to get to know TIM and to ask your questions to our experts during the live Q&A.