ML XGBoost Classification#484
Conversation
…other processes. Default to numerical index instead of string. (#478)
* `filter_spatial`: Clarified that a masking get applied for the given geometries. #469 * `filter_bbox`: Clarified that the bounding box is reprojected to the CRS of the spatial data cube dimensions if required. --------- Co-authored-by: Stefaan Lippens <soxofaan@users.noreply.github.com>
| "type": "object", | ||
| "subtype": "ml-model", | ||
| "title": "Machine Learning Model", | ||
| "description": "A machine learning model, accompanied with STAC metadata that implements the the STAC ml-model extension." |
There was a problem hiding this comment.
accompanied with STAC metadata that implements the the STAC ml-model extension
What does this practically mean here in this context of defining a JSON schema? Isn't that more a concern of a process like save_ml_model that actually "exports" the model to a more concrete form?
There was a problem hiding this comment.
accompanied with STAC metadata that implements the the STAC ml-model extension
What does this practically mean here in this context of defining a JSON schema? Isn't that more a concern of a process like
save_ml_modelthat actually "exports" the model to a more concrete form?
This was to remove the error due to the ml model being returned. I created a branch from the draft and not the ml branch.
| } | ||
| }, | ||
| { | ||
| "name": "seed", |
There was a problem hiding this comment.
a lot of parameters are listed here. Are these based on a particular xgboost implementation? And are we sure that they are translatable to other implementation? Otherwise I think it would make sense to drop a couple from this initial spec, and leave some wiggle room for backend implementers
I remember we had this same issue with the random forest process
There was a problem hiding this comment.
Yeah, I had checked python and R libraries and these params can be specified : https://xgboost.readthedocs.io/en/stable/parameter.html
I can mabye remove 3 : early_stopping_rounds, nfolds and nrounds to be dealt with internally. I set default values for most params though.
m-mohr
left a comment
m-mohr
left a comment
There was a problem hiding this comment.
@PondiB I think it would make sense to make PRs against the ml branch because otherwise all changes from the ML branch will also appear in this PR. This leads to confusion as you can see with @soxofaan's comments. Please rebase your changes against the ML branch if necessary and set the base branch of the PR to ml.
|
hmm this PR now has "27 files changed", most of the changes are irrelevant to the original issue (ML XGBoost) |
Lol, Never noticed it in the evening. I did run npm test locally no idea how it led to this. I will rebase the commit and push again. |
PondiB
force-pushed
the
ml-xgboost-class
branch
from
b83aaad to
e98dd7f
Compare
|
Moved to #487 |
3 participants
Specification for ML XGBoost for Classification.
Depends on #441