tensorflow / tensorflow/probability
Feature Request: Add crps method to tfp.distributions
Nobody has claimed this yet.
- Dominant language
- Jupyter Notebook
- Stars
- 4.4k
- Forks
- 1.1k
- PR merge metrics
- No merged PRs in 30d
Description
Hi,
first of all I'd like to thank you for the great work.
I have a feature request for a new method which implements the 'continuous ranked probability score' for all distributions (for which it is possible, see f.e. Evaluating Probabilistic Forecasts with scoringRules).
This method would enable the use of the CRPS as loss function in the same way as the negative log likelihood is used. (f.e. crps_loss_fn = lambda y, p_y : p_y.crps(y) )
Kind Regards
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing the tfp.distributions API and the linked paper on continuous ranked probability scores. Define the applicable distributions and expected crps(y) behavior, then add coverage so the method works wherever feasible and can be used as a loss function.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- tensorflow
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 20/100