Skip to content

chore: Adding more tests for On Demand Feature Views #4069

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
franciscojavierarceo merged 9 commits into from
Apr 4, 2024
Merged

chore: Adding more tests for On Demand Feature Views #4069

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
066880f
checking in progress...trying to fix tests
franciscojavierarceo Apr 3, 2024
79e6edc
testing more...
franciscojavierarceo Apr 3, 2024
262c586
fixed
franciscojavierarceo Apr 3, 2024
0ebfe7d
fixed some tests
franciscojavierarceo Apr 3, 2024
270f6c0
fixed test and serialization
franciscojavierarceo Apr 4, 2024
3574bcc
removed commented out code
franciscojavierarceo Apr 4, 2024
1d3745d
lint
franciscojavierarceo Apr 4, 2024
255ba05
added a test to make it explicit that feature calculation must happen…
franciscojavierarceo Apr 4, 2024
924ebd5
linter
franciscojavierarceo Apr 4, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
fixed test and serialization 
Signed-off-by: Francisco Javier Arceo <franciscojavierarceo@users.noreply.github.com>
  • Loading branch information
@franciscojavierarceo
franciscojavierarceo committed Apr 4, 2024
commit 270f6c0d30ac77c5f52deda9b3b62b682c730cef
13 changes: 13 additions & 0 deletions sdk/python/feast/on_demand_feature_view.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -300,10 +300,23 @@ def from_proto(
== "user_defined_function"
and on_demand_feature_view_proto.spec.feature_transformation.user_defined_function.body_text
!= ""
and on_demand_feature_view_proto.spec.mode == "pandas"
):
transformation = PandasTransformation.from_proto(
on_demand_feature_view_proto.spec.feature_transformation.user_defined_function
)
elif (
on_demand_feature_view_proto.spec.feature_transformation.WhichOneof(
"transformation"
)
== "user_defined_function"
and on_demand_feature_view_proto.spec.feature_transformation.user_defined_function.body_text
!= ""
and on_demand_feature_view_proto.spec.mode == "python"
):
transformation = PythonTransformation.from_proto(
on_demand_feature_view_proto.spec.feature_transformation.user_defined_function
)
elif (
on_demand_feature_view_proto.spec.feature_transformation.WhichOneof(
"transformation"
Expand Down
31 changes: 20 additions & 11 deletions sdk/python/tests/unit/test_on_demand_python_transformation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,6 @@

class TestOnDemandPythonTransformation(unittest.TestCase):
def setUp(self):
# data_dir = tempfile.TemporaryDirectory()
with tempfile.TemporaryDirectory() as data_dir:
self.store = FeatureStore(
config=RepoConfig(
Expand Down Expand Up @@ -83,10 +82,10 @@ def pandas_view(inputs: pd.DataFrame) -> pd.DataFrame:
mode="python",
)
def python_view(inputs: Dict[str, Any]) -> Dict[str, Any]:
Copy link
Copy Markdown
Collaborator

@tokoko tokoko Apr 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm still a little confused about the required signature here. Are these functions supposed to accept a dict of lists (looks like that in this test) and apply the udf for all entities at once? I thought from the previous PR that the goal was to have a udf that would be applied to individual entities...

Copy link
Copy Markdown
Collaborator

@tokoko tokoko Apr 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you also alter the tests so that more than one entity is passed? this will probably fail in such a case as only first entity is processed. If we are sticking with this signature, udf should look something like this:

return {
                'conv_rate_plus_acc_python': [
                    conv_rate + acc_rate
                    for conv_rate, acc_rate in zip(inputs['conv_rate'], inputs['acc_rate'])
                ]
            }

Copy link
Copy Markdown
Member Author

@franciscojavierarceo franciscojavierarceo Apr 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If you look at _infer_features_dict you'll see it expects a dict of lists. I added an explicit test that shows this will result in a type failure when running the apply operations. We can add singleton execution as a follow up but this is sufficient to highlight the currently supported behavior and then we can cut a release.

Copy link
Copy Markdown
Collaborator

@tokoko tokoko Apr 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@franciscojavierarceo got it, good... that's probably more efficient anyway. no rush, but in that case it will probably be a good idea to change type annotations for relevant functions to Dict[str, List[Any]].

Copy link
Copy Markdown
Member Author

@franciscojavierarceo franciscojavierarceo Apr 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I actually originally had that setup but I received a ton of type failures from that which is why I did it this way.

Let me address both of those as folllowups. I want to merge this and cut a release.

Copy link
Copy Markdown
Member Author

@franciscojavierarceo franciscojavierarceo Apr 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I made an issue here #4075, will close it later.

output: Dict[str, Any] = {
"conv_rate_plus_acc_python": inputs["conv_rate"]
+ inputs["acc_rate"]
}
output: Dict[str, Any] = {"conv_rate_plus_acc_python": []}
output["conv_rate_plus_acc_python"].append(
inputs["conv_rate"][0] + inputs["acc_rate"][0]
)
return output

self.store.apply(
Expand All @@ -110,7 +109,7 @@ def test_python_pandas_parity(self):
"driver_hourly_stats:acc_rate",
"python_view:conv_rate_plus_acc_python",
],
).to_df()
).to_dict()

online_pandas_response = self.store.get_online_features(
entity_rows=entity_rows,
Expand All @@ -121,10 +120,20 @@ def test_python_pandas_parity(self):
],
).to_df()

assert len(online_python_response) == 4
assert all(
key in online_python_response.keys()
for key in [
"driver_id",
"acc_rate",
"conv_rate",
"conv_rate_plus_acc_python",
]
)
assert len(online_python_response["conv_rate_plus_acc_python"]) == 1
assert (
online_python_response["conv_rate_plus_acc_python"]
.equals(online_pandas_response["conv_rate_plus_acc_pandas"])
.equals(
online_python_response["conv_rate"] + online_python_response["acc_rate"]
)
online_python_response["conv_rate_plus_acc_python"][0]
== online_pandas_response["conv_rate_plus_acc_pandas"][0]
== online_python_response["conv_rate"][0]
+ online_python_response["acc_rate"][0]
)