End of course summary
Understanding, evaluating, and improving forecasts of infectious disease burden
Learning objectives
- improve understanding of forecasting as an epidemiological task
- familiarity with metrics for evaluating probabilistic forecasts and their properties
- ability to score probabilistic forecasts in R
- understanding of predictive ensembles and their properties
- ability to create a predictive ensemble of forecasts in R
- understanding the concept of weighted forecast ensembles
Timeline
- forecast visualisation (session 1)
- forecast evaluation (session 2)
- forecast evaluation of multiple models (session 3)
- forecast ensembles (session 4)
Key takeaways
Session 1: Forecasting
- forecasting is the task of making unconditional statements about the future
- meaningful forecasts are probabilistic
- we can use visualisation understand the predictive performance of a model
Session 2: Forecast evaluation
- we can assess forecasts using proper scoring rules
- we can use scoring to quantify the predictive performance of different models
Session 3: Forecast ensembles
- ensembles can combine forecasts from multiple models
- simple ensembles often outperform indivdiual models
- weighted ensembles can learn from past performance aiming to make better forecasts
Outlook
- it is worth trying some of these methods here in practice to learn more about typical forecast performance
- one way of doing so is by contributing to forecast hubs
Further learning
- Some suggested reading is given in the End of course summary and discussion
- You can find broader course material on the topic at https://nfidd.github.io/nfidd
Feedback
- please tell us if you enjoyed the course, what worked / didn’t work etc.
- we will send out a survey for feedback
Thank you for attending!
End of course summary