Update split_data with new default for timeseries #3650
Conversation
Codecov Report
@@ Coverage Diff @@
## main #3650 +/- ##
=======================================
+ Coverage 99.7% 99.7% +0.1%
=======================================
Files 335 335
Lines 33790 33840 +50
=======================================
+ Hits 33662 33712 +50
Misses 128 128
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
eccabay
requested review from
jeremyliweishih,
chukarsten,
christopherbunn,
fjlanasa and
MichaelFu512
and removed request for
jeremyliweishih
| @@ -2,12 +2,13 @@ Release Notes | |||
| ------------- | |||
| **Future Releases** | |||
| * Enhancements | |||
| * Add ``exclude_featurizers`` parameter to ``AutoMLSearch`` to specify featurizers that should be excluded from all pipelines :pr:`3631` | |||
| * Added ``exclude_featurizers`` parameter to ``AutoMLSearch`` to specify featurizers that should be excluded from all pipelines :pr:`3631` | |||
There was a problem hiding this comment.
Make sure you send a picture of this to @angela97lin
| if is_binary(problem_type): | ||
| X, y = X_y_binary | ||
| if is_multiclass(problem_type): | ||
| X, y = X_y_multi | ||
| if is_regression(problem_type): | ||
| X, y = X_y_regression | ||
| problem_configuration = None | ||
| if is_time_series(problem_type): | ||
| problem_configuration = {"gap": 1, "max_delay": 7, "time_index": "ts_data"} | ||
|
|
||
| X = make_data_type(data_type, X) | ||
| y = make_data_type(data_type, y) |
There was a problem hiding this comment.
I suspect this is boiler plate not just for these two tests but also across many of the tests. Did you want to refactor it into something that returns problem_configuration, X, and y given data_type, problem_type, etc? I am pretty sure I've seen this a few other places too. I think it's also fine to file an issue to call out the refactor with the modules we should refactor this code into.
There was a problem hiding this comment.
I like this. A slight refactor to get_test_data_from_configuration works wonderfully to factor this out! We can go through and replace other instances in a future story.
Reduces the default
test_sizeinsplit_datato be 0.1 instead of 0.2, as long as the new value is greater than the passed in forecast horizon.