score_df <- neon4cast::combined_scores(theme = "phenology")
class_model_ids <- #set of team names for students in class
score_df <- score_df |> filter(model_id %in% class_model_ids)8 Using the Challenge in Classes
The NEON Ecological Forecasting Challenge is a powerful resource to use in teaching. We encourage courses to have student submit to the challenge as projects in traditional courses or training workshops.
If you are interested in using the Challenge in a class or workshop, we welcome you to reach out to us at eco4cast.initiative@gmail.com
NEON wrote a blog post about using the Challenge in courses here
Based the experiences of 7+ courses using the Challenge, we have an evolving set of best practices
- Consider data lags: The Phenology, Terrestrial, and Aquatic challenges have data available in near-real time, therefore students submitting to these Challenges will multiple opportunities to see how their forecasts perform: Phenology is 1-day, Terrestrial is 5-days, and Aquatics is 2-days. In contrast the Beetles and Ticks challenges have 60 - 300 day latency so students may not see their forecasts evaluated in the workshop or semester.
- Add meaning: Have students pick sites that interest them. There is not expectation that students submit forecasts for all sites in a theme.
- Start simple: Encourage students to start with simple models.
- Normalize imperfection: Work with students to embrace that all models are wrong, and putting the forecast out there is the key step.
- Use the example models as a template: have students build their models from the simple example models that have been made available Appendix A is one example and ?sec-nulls are the nulls submitted to the challenge.
- Point students to documentation: the useful functions (?sec-functions) and FAQ (Chapter 10)
- Reach out: Reach out to the Challenge organizers for questions when planning your course or workshop (eco4cast.initiative@gmail.com)
If you want to access students scored forecasts, the following R function in the neon4cast package can be used (where the argument is the theme name):