Investigate methods for notifying research team (here, Galaxy Zoo team) and Zoo team in the case that:
- Most primary (not spice / random) subjects placed in target subject set are retired and no longer getting served to volunteers.
- The motivation: when most target subjects are retired, it is time for the research team to evaluate whether to continue classifying the current subject set or not.
- The completeness fraction for the target workflow on the subject set resource could be checked programmatically -- when greater than 0.95 (or similar), notify project team or flag to Zoo team.
- Predictions from the model change significantly from week to week.
2. The motivation: some results from the weekly runs are available to the research team, but it would be helpful to have "push" notifications if model predictions change significantly -- meaning that the project team should take a look and potentially change how training is run.
Investigate methods for notifying research team (here, Galaxy Zoo team) and Zoo team in the case that:
2. The motivation: some results from the weekly runs are available to the research team, but it would be helpful to have "push" notifications if model predictions change significantly -- meaning that the project team should take a look and potentially change how training is run.