Export model selector defaults + metadata fixes #199
Conversation
…ti, regression) 2. Skip unsupported metadata values when producing model selector summary metadata
Codecov Report
@@ Coverage Diff @@
## master #199 +/- ##
===========================================
- Coverage 86.4% 69.14% -17.26%
===========================================
Files 309 310 +1
Lines 10009 10018 +9
Branches 351 526 +175
===========================================
- Hits 8648 6927 -1721
- Misses 1361 3091 +1730
Continue to review full report at Codecov.
|
Codecov Report
@@ Coverage Diff @@
## master #199 +/- ##
=======================================
Coverage 86.38% 86.38%
=======================================
Files 310 310
Lines 10019 10019
Branches 550 550
=======================================
Hits 8655 8655
Misses 1364 1364Continue to review full report at Codecov.
|
…model-selector-improvements
…e/TransmogrifAI into mt/model-selector-improvements
|
@kinfaikan please review |
| } | ||
| // if models to use has been specified and the models have been specified - filter the models by the names | ||
| else if ( | ||
| modelTypesToUse.distinct.sortBy(_.entryName) != modelDefaults.modelTypesToUse.distinct.sortBy(_.entryName) |
There was a problem hiding this comment.
Why not simply use entire modelsAndParameters?
There was a problem hiding this comment.
well, because it would yield incorrect results: List(1,2) == List(2,1) // false. That the reason I am dedupping and sorting here.
There was a problem hiding this comment.
I was thinking whether it made more sense to ignore modelTypes and modelDefaults when modelsAndParameters was not empty.
Related issues
Describe the proposed solution
Describe alternatives you've considered
N/A