Skip to content

Feat: local template odfv write #6892

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

Open
ryota717 wants to merge 5 commits into
base: master
Choose a base branch
from
Open

Feat: local template odfv write #6892

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
808aa05
feat: Add native Python mode on demand feature view to the local temp…
ryota717 Sep 29, 2026
711a9d4
fix: Create sqlite tables for on demand feature views with write_to_o…
ryota717 Sep 29, 2026
eb20131
feat: Add transform-on-write on demand feature view to the local temp…
ryota717 Sep 29, 2026
736e005
test: Verify transform-on-write features are materialized by the loca…
ryota717 Sep 29, 2026
d2c15f8
docs: Describe the on demand feature views added to the local template
ryota717 Sep 29, 2026
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
Next Next commit
feat: Add native Python mode on demand feature view to the local temp… 
…late

The default template already demonstrates a Pandas mode on demand feature
view. Add transformed_conv_rate_python, the same transformation written
in native Python mode (mode="python"), so that feast init shows both
transformation modes side by side, and fetch its features in
test_workflow.py during both historical and online retrieval.

Part of #5478

Signed-off-by: ryota717 <ryota.nishijima@fout.jp>
  • Loading branch information
@ryota717
ryota717 committed Sep 29, 2026
commit 808aa059fb084d7eb170b4a93389814d0adffb0f
35 changes: 34 additions & 1 deletion sdk/python/feast/templates/local/feature_repo/feature_definitions.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
# This is an example feature definition file

from datetime import timedelta
from typing import Any

import pandas as pd

Expand Down Expand Up @@ -89,7 +90,8 @@


# Define an on demand feature view which can generate new features based on
# existing feature views and RequestSource features
# existing feature views and RequestSource features. By default the transformation
# runs in Pandas mode (mode="pandas"): the UDF receives and returns a DataFrame.
@on_demand_feature_view(
sources=[driver_stats_fv, input_request],
schema=[
Expand All @@ -104,6 +106,37 @@ def transformed_conv_rate(inputs: pd.DataFrame) -> pd.DataFrame:
return df


# The same transformation written in native Python mode (mode="python"). The UDF
# receives a dict mapping each input feature name to a list of values (one per
# row) and returns a dict with the same shape. This avoids the Pandas overhead
# for small online requests and is often easier to reason about.
#
# Only the features the UDF needs are selected from the source feature view.
# This is required here: driver_stats_fv also has Map / Struct / Json fields, and
# Python mode feature inference cannot generate sample values for those types.
@on_demand_feature_view(
sources=[driver_stats_fv[["conv_rate"]], input_request],
schema=[
Field(name="conv_rate_plus_val1_python", dtype=Float64),
Field(name="conv_rate_plus_val2_python", dtype=Float64),
],
mode="python",
)
def transformed_conv_rate_python(inputs: dict[str, Any]) -> dict[str, Any]:
return {
"conv_rate_plus_val1_python": [
conv_rate + val_to_add
for conv_rate, val_to_add in zip(inputs["conv_rate"], inputs["val_to_add"])
],
"conv_rate_plus_val2_python": [
conv_rate + val_to_add_2
for conv_rate, val_to_add_2 in zip(
inputs["conv_rate"], inputs["val_to_add_2"]
)
],
}


# This groups features into a model version
driver_activity_v1 = FeatureService(
name="driver_activity_v1",
Expand Down
8 changes: 8 additions & 0 deletions sdk/python/feast/templates/local/feature_repo/test_workflow.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,8 +93,12 @@ def fetch_historical_features_entity_df(store: FeatureStore, for_batch_scoring:
"driver_hourly_stats:conv_rate",
"driver_hourly_stats:acc_rate",
"driver_hourly_stats:avg_daily_trips",
# Pandas mode on demand transformation
"transformed_conv_rate:conv_rate_plus_val1",
"transformed_conv_rate:conv_rate_plus_val2",
# Native Python mode on demand transformation
"transformed_conv_rate_python:conv_rate_plus_val1_python",
"transformed_conv_rate_python:conv_rate_plus_val2_python",
],
).to_df()
print(training_df.head())
Expand Down Expand Up @@ -124,8 +128,12 @@ def fetch_online_features(store, source: str = ""):
"driver_hourly_stats:driver_metadata",
"driver_hourly_stats:driver_config",
"driver_hourly_stats:driver_profile",
# Pandas mode on demand transformation
"transformed_conv_rate:conv_rate_plus_val1",
"transformed_conv_rate:conv_rate_plus_val2",
# Native Python mode on demand transformation
"transformed_conv_rate_python:conv_rate_plus_val1_python",
"transformed_conv_rate_python:conv_rate_plus_val2_python",
]
returned_features = store.get_online_features(
features=features_to_fetch,
Expand Down