9/23/2021 Update #408
isaacmg
announced in
Announcements
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
We continue to move forward towards Flow Forecast (FF) release 1.0 and plan on hopefully rolling it out sometime in early October. We would like to close out the following issues before the release in particular:
Besides the above issues we also want to add informative docstrings to all of our major FF functions and models. Additionally, pages on Confluence should be created with lists of all supported models, loss functions, and embeddings. Finally, we want to organize all of our FF notebooks in an easy to use format. Right now there are many interesting forecasting notebooks on a variety of TS problems but they are hard to find and navigate. Whether to include series_id_support is another major consideration. As it will likely require a lot of additional unit-tests. What do you think? Comment below.
Aside from the rollout of 1.0 we are looking forward to another active Hacktoberfest 2021 . We hope to recruit many new contributors and users to FF. Whether you are a university undergrad getting your feet wet or a seasoned deep learning researcher FF can use your help. We have tons of issues and our core maintainers are more than willing to help get your started on a PR! Please join and contribute.
Time Series Classification is now fully supported in FF. However, there are still many kinks to iron and additional features. Most of these features like a focal loss function (for anomaly detection), LabelEncoder as a parameter, and multitask forecasting and classification will likely come in 1.01+ releases though.
Do you have an example notebook or interesting problem you used FF to solve? Feel free to post a link below or describe your use-case.
Beta Was this translation helpful? Give feedback.
All reactions