Ad Hoc Runs and Copy HPOs

Myst Platform Release (2022-08-11)

Hello Myst Platform users,

We have two new enhancements this week. You can now run model fits and time series on an ad hoc basis and copy HPOs from the UI.

Please update your myst-alpha package at your earliest convenience to ensure you have access to this week’s updates (instructions here).

⚡️Enhancements:

• You can now run fits and runs time series on an ad hoc basis! This feature is available both in the web application and client library. See our documentation on creating ad hoc results for more information.
• You can now copy HPO runs from the UI. From the HPO table in the experiments tab, select the three dots to view the copy HPO button.

✅ Bug fix:

• We have adjusted an edge case in the resampling operation using interpolation that previously caused issues when running on data up to the latest available time period. If you are using resampling with interpolation where values exist beyond the last value being resampled, we previously would use the “end_time + 1” value for interpolation. Now, we will instead flat line and pad the upsampling for the final increment. For example, let’s consider a time series with values 10, 20, 30 at hours 1, 2, and 3. If we feed this time series into a resampling connector and apply a run policy to its output for hours 1 until 3 (recalling that Myst is not end-time-inclusive), previously you would have gotten (10, 12.5, 15, 17.5, 20, 22.5, 25, 27.5). Now you would get (10, 12.5, 15, 17.5, 20, 20, 20, 20). This small tweak allows seamless resampling on data all the way up until the last available time period, where before such interpolation would have required an as-yet-nonexistent future measurement.

Thank you!
Charlie and the Myst team