Remove ComponentGraph.from_list and update PipelineBase impl creating from lists
#2549
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov Report
@@ Coverage Diff @@
## main #2549 +/- ##
=======================================
+ Coverage 99.9% 99.9% +0.1%
=======================================
Files 285 285
Lines 26167 26168 +1
=======================================
+ Hits 26131 26132 +1
Misses 36 36
Continue to review full report at Codecov.
|
angela97lin
commented
Jul 26, 2021
| @@ -31,6 +31,10 @@ class ComponentGraph: | |||
| def __init__(self, component_dict=None, random_seed=0): | |||
| self.random_seed = random_seed | |||
| self.component_dict = component_dict or {} | |||
| if not isinstance(self.component_dict, dict): | |||
There was a problem hiding this comment.
No more lists allowed!
angela97lin
commented
Jul 26, 2021
| @@ -150,6 +151,34 @@ def linearized_component_graph(self): | |||
| """A component graph in list form. Note that this is not guaranteed to be in proper component computation order""" | |||
| return ComponentGraph.linearized_component_graph(self.component_graph) | |||
|
|
|||
| @staticmethod | |||
| def _make_component_dict_from_component_list(component_list): | |||
There was a problem hiding this comment.
Moved this into pipeline_base... mostly because trying to import this from pipeline.utils.py, which imports PipelineBase, would otherwise create a circular dependency :/
angela97lin
requested review from
dsherry,
chukarsten,
freddyaboulton,
bchen1116,
jeremyliweishih and
ParthivNaresh
evalml/pipelines/pipeline_base.py
Outdated
| component = handle_component_class(component) | ||
| if not component._supported_by_list_API: | ||
| raise ValueError( | ||
| "This component cannot be defined in a list because edges may be ambiguous. Please use a dictionary to specify the appropriate component graph for this pipeline instead." |
There was a problem hiding this comment.
nit-pick: Should we list the component name in the message?
There was a problem hiding this comment.
Sure thing, done 😁
freddyaboulton
approved these changes
Jul 26, 2021
| "SMOTENC Oversampler", | ||
| estimator, | ||
| ] | ||
| component_graph={ |
This was referenced Jul 26, 2021
Merged
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
More breaking up #2490 into smaller and cleaner PRs :)
This PR removes
from_listfrom ComponentGraph and updatesPipelineBase's implementation to handle creating a component dictionary to pass along to the ComponentGraph.This is slightly different from our behavior today because it puts the burden of converting from a list of components --> component dict on the pipeline. This means that ComponentGraphs do not need to understand how to handle lists.
Note that
_make_component_dict_from_component_listupdates the generated component graph to include X and y edges for each component, but it is not enforced yet! That will come in a later PR :)