FluxML
[WIP] Blog about working with Time Series Data using FastAI.jl
This blog demonstrates the work done for adding time series support to FastAI.jl as part of GSoC 22. It contains basic sample code and gives an overview of currently supported features under this submodule. Will add more stuff in the future. This was done under the mentorship of @darsnack @ToucheSir @lorenzoh .
Do you mind writing up a quick list of additions you have planned? That way we're not suggesting stuff you're already working on :)
Primally I was looking to add these :-
- More information about the models (diagram, blocks, etc)
- Examples of the model working (classifying a sample time series)
Would love to hear suggestions on how I could improve this.
I left a few comments that do not need to be addressed before the GSoC deadline, but they should be taken care of before merging.
The main thing that needs to be done before GSoC is to add some notes about future work, and also include links to the PRs somewhere in the post.
For GSoC, you can preview this page locally and save it as a PDF. Upload that PDF somewhere and create a permalink to it (e.g. Google Drive). This is how Google prefers the submission.
Will resolve the comments you guys made. For GSoC I was planning to submit this gist. It contains the PRs along with the blogs attached, similar to what manikya did last year.
Generally LGTM. The biggest addition I think would be providing some visual of what the data and predictions look like, similar to what we've talked about for the notebooks.
Yeah that is in the works, will hopefully add soon.