Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Time series gap detection for TFT tasks #754

Open
gmdiana-hershey opened this issue Oct 10, 2022 · 1 comment · May be fixed by #770 or #878
Open

Time series gap detection for TFT tasks #754

gmdiana-hershey opened this issue Oct 10, 2022 · 1 comment · May be fixed by #770 or #878
Assignees
@int-chaos

Comments

@gmdiana-hershey
Copy link
Collaborator

I've been getting a warning when I'm using TemporalFusionTransformer, indicating missing timestamps. It's coming from

FLAML/flaml/automl.py

Line 1033 in f97c233

"Missing timestamps detected. To avoid error with estimators, set estimator list to ['prophet']. "
. It Seems like perhaps it's checking the whole dataset as a single time series, rather than each time series individually. I think that adding logic to _validate_ts_data() to check if the task is 'ts_forecast_panel' and using the group_ids passed when fit() is called could be a solution.
@int-chaos
Copy link
Collaborator

int-chaos commented Oct 12, 2022 •
edited
Loading

if you think you have an idea on how to fix it, feel free to attempt it and submit a pr. FYI, group_ids is already passed into fit() and you can get it via self._state.fit_kwargs.get("group_ids").

but otherwise, I can try what I have in mind (basically separate the entire time series by groups and check for each group).

@int-chaos int-chaos linked a pull request Oct 19, 2022 that will close this issue
Open
@sonichi sonichi linked a pull request Oct 19, 2022 that will close this issue
tft gap detection on each time series independently #770
Open
@int-chaos int-chaos linked a pull request Jan 9, 2023 that will close this issue
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@int-chaos int-chaos

Labels
None yet
Projects
None yet
Milestone
No milestone
2 participants
@gmdiana-hershey @int-chaos