Skip to content

docs(odfv): document column naming behavior when using multiple FeatureView sources #6087

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
anix-lynch wants to merge 1 commit into
base: master
Choose a base branch
from
Open

docs(odfv): document column naming behavior when using multiple FeatureView sources #6087

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
29 changes: 29 additions & 0 deletions docs/reference/beta-on-demand-feature-view.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -303,6 +303,35 @@ This approach allows for a hybrid workflow where you can:

Even when features are materialized with transformations skipped (`transform_on_write=False`), the feature server can still apply transformations during API calls for any missing values or for features that require real-time computation.

## Column Naming in the Input DataFrame

When writing transformation functions, the column names available in the input DataFrame depend on the source types used:

| Source Type | Column name format | Example |
|---|---|---|
| `RequestSource` | feature name only | `features_df["val_to_add"]` |
| Single `FeatureView` source | feature name only | `features_df["conv_rate"]` |
| Multiple `FeatureView` sources | `{feature_view_name}__{feature_name}` | `features_df["driver_hourly_stats__conv_rate"]` |

When more than one `FeatureView` is passed as a source, Feast prefixes each column with the feature view name and double underscore (`__`) to avoid naming conflicts. This prefix is **not added** for `RequestSource` columns or when only a single `FeatureView` is used.

**Defensive access pattern** when the number of FeatureView sources may vary:

```python
@on_demand_feature_view(
sources=[driver_hourly_stats_view, driver_profile_view],
schema=[Field(name="combined_score", dtype=Float64)],
mode="pandas",
)
def combined_driver_score(features_df: pd.DataFrame) -> pd.DataFrame:
# Handles both single-source (no prefix) and multi-source (prefixed) cases
col = "driver_hourly_stats__conv_rate"
conv_rate = features_df[col] if col in features_df.columns else features_df["conv_rate"]
df = pd.DataFrame()
df["combined_score"] = conv_rate * 100
return df
```

## CLI Commands
There are new CLI commands to manage on demand feature views:

Expand Down