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.

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

Address issues #27, #28, #29, #34 about timestamps & sampling granularities. #30

Merged
merged 15 commits into from
Oct 18, 2021

Conversation

aadyotb
Copy link
Contributor

@aadyotb aadyotb commented Oct 11, 2021

Issues #27, #28, #29, and #34 point out some limitations on how Merlion handles monthly/quarterly/yearly sampling granularities. Specifically, we assume that all regular sampling granularities must be a fixed number of seconds, but this is not true for commonly used granularities like monthly or quarterly.

This PR changes the internal representation of timedeltas to be pandas objects/strings in various places, rather than a float number of seconds. This addresses Issues #29 and #34. It also fixes a bug where ForecasterBase.resample_time_stamps mishandled an integer number of time stamps (#28) and updates Sarima's default granularity to be None rather than 1h (addressing #27).

Fix the handling of integer # of timestamps in resample_time_stamps.
Make SARIMA resample to default seasonality, instead of hourly.
@aadyotb aadyotb requested a review from paulkass October 12, 2021 20:19
@aadyotb
Copy link
Contributor Author

aadyotb commented Oct 12, 2021

@datenzauberai this PR should address the issues you raised. Please let me know if you have further concerns.

@datenzauberai
Copy link

Awesome that you’ve been able to make this big change this fast! Looking forward to try this out.

@julioasotodv
Copy link

Thank you @aadyotb for tackling this issue! Since a lot of time series out in the wild are monthly-based, right now most models in Merlion are unusable in a lot of scenarios.

@aadyotb aadyotb marked this pull request as ready for review October 14, 2021 21:46
@aadyotb aadyotb changed the title Address issues #27, #28, #29 about timestamps & sampling granularities. Address issues #27, #28, #29, #34 about timestamps & sampling granularities. Oct 15, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

4 participants