Repository navigation
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #871 +/- ##
=======================================
Coverage 99.68% 99.68%
=======================================
Files 201 202 +1
Lines 38176 38245 +69
Branches 2442 2448 +6
=======================================
+ Hits 38057 38126 +69
Misses 51 51
Partials 68 68
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
|
Seem to be some style issues that need to be addressed. I'm still trying to fully understand this issue and to think about what the correct way forward is. It may be that the right way is to raise some sort of ambiguity error in the case of ambiguous positional arguments. |
|
@bashtage pushed the formatting fixes. Black, isort, Ruff and flake8 pass locally. I've left the positional-input behavior unchanged for now and will wait for your direction on the ambiguity error. |
Positional arrays can retain generated entity/time groups that disagree with the dependent data. This can reverse
FirstDifferenceOLScoefficient signs and misgroup within/between transformations or weights.Use the dependent index for validated, positionally paired regressors and weights before transformations. Added coverage for ordered/shuffled rows, missing observations, labelled/array inputs and weighted models.
The new regressions produce 33 baseline failures. Final panel tests pass 6,858 cases with 367 skips and 47 expected failures. Independent chronological and weighted least-squares checks pass 627 tests. Ruff passes.
Validation used the pure-Python runtime with slow tests excluded. Native/Cython, wheel, full-repository and additional-platform validation remain pending.
Prepared with AI assistance.