Explainable AI as Uncertainty in Forecasting

Forecasts are done to make decisions about the future. When you know it will rain, you take an umbrella. But what if there is only a chance? How do you express chance numerically? In this example, weather forecasting apps often already include a probability with their predictions. Many other applications however do not.
Do you think it would be useful if chatGPT could tell you how sure it is about its statements?
 
 Uncertainty in forecasting, or probabilistic forecasting, is a hot topic within the machine learning- and many adjacent communities. Instead of giving a seemingly precise prediction, we instead five a range of possible outcomes. This is more informative and can be used to make better decisions. Moreover, it makes machine learning models more transparent and understandable.

There exists a plethora of ways to express uncertainty in forecasting. One big challenge here is that there are so many methods, and it is not clear which one is the best (or for what reason). In this thesis project, you will compare different methods for expressing uncertainty in forecasting. You will evaluate the methods on their approaches and compare their performance.