Add multiseries time series regression pipeline#4256
Conversation
Codecov Report
@@ Coverage Diff @@
## main #4256 +/- ##
=======================================
+ Coverage 99.7% 99.7% +0.1%
=======================================
Files 351 353 +2
Lines 38512 38643 +131
=======================================
+ Hits 38391 38522 +131
Misses 121 121
|
eccabay
requested review from
MichaelFu512,
christopherbunn,
chukarsten,
jeremyliweishih and
remyogasawara
jeremyliweishih
left a comment
jeremyliweishih
left a comment
There was a problem hiding this comment.
Overall LGTM but requesting one small test case and had some clarifying questions!
evalml/tests/pipeline_tests/regression_pipeline_tests/test_multiseries_regression_pipeline.py
Show resolved
Hide resolved
christopherbunn
left a comment
christopherbunn
left a comment
There was a problem hiding this comment.
One question but otherwise LGTM
| return stacked_series | ||
|
|
||
|
|
||
| def stack_X(X, series_id_name, time_index, starting_index=None): |
There was a problem hiding this comment.
Just to double check, this is able to handle exogenous features too right? (e.g. stacking columns feature_a, feature_b, and feature_c columns into one feature column). If so, is there a test for this?
There was a problem hiding this comment.
Correct. stack_data will stack a dataframe of a single unstacked column, and stack_X is essentially just a wrapper around stack_data which calls it for every original stacked column. The test is test_stack_X in the pipeline_utils tests
chukarsten
left a comment
chukarsten
left a comment
There was a problem hiding this comment.
Sorry, just a few nitty things!
| return self | ||
|
|
||
| def _fit(self, X, y): | ||
| from evalml.pipelines.utils import unstack_multiseries |
There was a problem hiding this comment.
Let's maybe move this to the top with the other imports?
There was a problem hiding this comment.
Circular imports if moved to the top - I'm open to any other suggestions on how to mitigate!
There was a problem hiding this comment.
Can we file a ticket to deal with that? That makes me...uncomfortable.
| return self | ||
|
|
||
| def _fit(self, X, y): | ||
| from evalml.pipelines.utils import unstack_multiseries |
There was a problem hiding this comment.
Can we file a ticket to deal with that? That makes me...uncomfortable.
4 participants
Closes #4254