add dim_reduce to aggregates, refactor patch utils #456
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #456 +/- ##
========================================
Coverage 99.84% 99.85%
========================================
Files 114 115 +1
Lines 9258 9372 +114
========================================
+ Hits 9244 9358 +114
Misses 14 14
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
dascore/proc/aggregate.py
Outdated
| dim_reduce | ||
| How to reduce the dimensional coordinate associated with the | ||
| aggregated axis. Can be the name of any valid aggregator, a callable, | ||
| "empty" - which returns and empty coord, or "squeeze" which drops |
There was a problem hiding this comment.
It would be helpful to note that "empty" is the default behavior.
| dim_reduce: str | Callable = "empty", | ||
| ) -> PatchType: | ||
| """ | ||
| Aggregate values along a specified dimension. |
There was a problem hiding this comment.
The notes section in the docs should be revised based on the new dim_reduce argument.
| is found in args, its corresponding values is `None`. | ||
| Return the name of the dimension, its axis position, and its value. | ||
| Examples |
There was a problem hiding this comment.
Thanks for including these examples in the docs. Super helpful as this function is being used in a lot of other functions.
ahmadtourei
left a comment
ahmadtourei
left a comment
There was a problem hiding this comment.
Looks great! Please see my comments on a couple of suggestions.
Description
This PR does 3 sorta related things:
Adds a
dim_reducekeyword toPatch.aggregateand friends (eg mean, min, max) which controls what happens to the aggregated dimension. Options now are to apply another numpy-like reducer, squeeze out the dimension, or empty it. The default behavior is unchanged, but the additional options are useful when calculating stats over patches in a spool.Unified getting dimension name, axis, and values from args and kwargs into one new function
dascore.utils.patch.get_dim_axis_valueand replaced usages of two overlapping functions that did this before.Added a
validate_calloption topatch_functionwhich uses pydantic's validate_call decorator to do type checking on patch functions when enabled. This makes it much easier to enforce certain types and ranges on patch function parameters.Checklist
I have (if applicable):