search

lanl / SEPIA

Simulation-Enabled Prediction, Inference, and Analysis: physics-informed statistical learning.
Other
33 stars 7 forks source link

transform from standardized to native method #32

Open jgattiker opened 3 years ago

jgattiker commented 3 years ago

A utility function in SepiaData to transform from the unit cube to the native scale would be handy. Specific case: x's and t's for e.g. prediction are supplied to Predict on the native scale. Sampling randomly in the domain is more natural on unit cube. Should be able to transform these out to native without digging into the data structures.

luiarthur commented 3 years ago

Can the MinMaxScaler be used here?

jgattiker commented 3 years ago

Probably. We do scaling and keep track of the scaling parameters in data. This is really just a post-it to remind me to do this sometime, after doing it as a user would (inconveniently).

I hope some additional funding to support sepia development next FY will come through, and can polish it up a bit - including testing.

I just made some changes without branching/merging (relying on the test cases running locally, and things working on a calibration problem I'm looking at). I don't know how to do that yet, I still need to get comfortable with that.

On Thu, Aug 12, 2021 at 11:43 AM Arthur Lui @.***> wrote:

Can the MinMaxScaler https://scikit-learn.org/stable/modules/preprocessing.html#scaling-features-to-a-range be used here?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/lanl/SEPIA/issues/32#issuecomment-897834895, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABNMXTFCATWMKORDNYYPIX3T4QB2ZANCNFSM5CBXICNQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email .