Unable to register the feature view and on demand feature view together

[itsmano1993]

Code snippet

import pandas as pd
from feast import FeatureStore, FeatureView, Field, FileSource, Entity, FeatureService, OnDemandFeatureView, \
    RequestSource
from feast.on_demand_feature_view import on_demand_feature_view
from feast.transformation.python_transformation import PythonTransformation
from feast.types import Int64, Float32
import datetime

store = FeatureStore(".")

customer_data_source = FileSource(
    path="data/customer_features.parquet",  # Can be a local file, S3, GCS, etc.
    timestamp_field="event_timestamp",  # Required for time-series data
    created_timestamp_column="created_timestamp"
)

customer = Entity(name="customer", join_keys=["customer_id"])

customer_features_view = FeatureView(name="customer_features", source=customer_data_source, entities=[customer],
                                     ttl=datetime.timedelta(days=365),
                                     schema=[
                                         Field(name="age", dtype=Int64),
                                         Field(name="income", dtype=Float32),
                                     ])

input_request = RequestSource(
    name="vals_to_add",
    schema=[
        Field(name="val_to_add", dtype=Int64)
    ],
)


@on_demand_feature_view(
    sources=[customer_features_view, input_request], schema=[
        Field(name="age_income_ratio", dtype=Float32),
    ], mode="pandas", write_to_online_store=True)
def calculate_ondemand_features(df: pd.DataFrame) -> pd.DataFrame:
    df["age_income_ratio"] = df["age"] / df["income"]
    return df


customer_feature_service = FeatureService(name="customer_feature_service", features=[customer_features_view, calculate_ondemand_features])

store.apply([customer, customer_features_view, customer_feature_service, calculate_ondemand_features])

store.materialize(
    start_date=datetime.datetime(2024, 1, 1),  # Historical start date
    end_date=datetime.datetime.now()  # Latest available data
)

Error details

feast.errors.SpecifiedFeaturesNotPresentError: Explicitly specified features [Field(
    name='age_income_ratio',
    dtype=<PrimitiveFeastType.FLOAT32: 6>,
    description='',
    tags={}
    vector_index=False
    vector_search_metric=None
)] not found in inferred list of features [Field(
    name='customer_features__age',
    dtype=<PrimitiveFeastType.INT64: 4>,
    description='',
    tags={}
    vector_index=False
    vector_search_metric=None
), Field(
    name='age',
    dtype=<PrimitiveFeastType.INT64: 4>,
    description='',
    tags={}
    vector_index=False
    vector_search_metric=None
), Field(
    name='customer_features__income',
    dtype=<PrimitiveFeastType.FLOAT64: 5>,
    description='',
    tags={}
    vector_index=False
    vector_search_metric=None
), Field(
    name='income',
    dtype=<PrimitiveFeastType.FLOAT64: 5>,
    description='',
    tags={}
    vector_index=False
    vector_search_metric=None
), Field(
    name='val_to_add',
    dtype=<PrimitiveFeastType.INT64: 4>,
    description='',
    tags={}
    vector_index=False
    vector_search_metric=None
), Field(
    name='age_income_ratio',
    dtype=<PrimitiveFeastType.FLOAT64: 5>,
    description='',
    tags={}
    vector_index=False
    vector_search_metric=None
)] for 'calculate_ondemand_features'

Dataset creation

import pandas as pd
import datetime
import os

# Ensure the data directory exists
# os.makedirs("data", exist_ok=True)

# Create sample data
data = {
    "customer_id": [101, 102, 103, 104, 105],  # Entity IDs
    "age": [25, 34, 45, 29, 40],  # Age feature
    "income": [50000.0, 72000.5, 83000.0, 62000.7, 91000.3],  # Income feature
    "event_timestamp": [
        datetime.datetime(2024, 1, 1, 10, 0),
        datetime.datetime(2024, 1, 2, 12, 0),
        datetime.datetime(2024, 1, 3, 14, 0),
        datetime.datetime(2024, 1, 4, 16, 0),
        datetime.datetime(2024, 1, 5, 18, 0),
    ],
    "created_timestamp": [
        datetime.datetime(2024, 1, 1, 9, 30),
        datetime.datetime(2024, 1, 2, 11, 45),
        datetime.datetime(2024, 1, 3, 13, 15),
        datetime.datetime(2024, 1, 4, 15, 20),
        datetime.datetime(2024, 1, 5, 17, 10),
    ],
}

# Convert to DataFrame
df = pd.DataFrame(data)

# Save as Parquet
df.to_parquet("data/customer_features.parquet", index=False)

print("Parquet file 'customer_features.parquet' generated successfully!")

[franciscojavierarceo]

which version of feast are you using?

[franciscojavierarceo]

can you share your feature_store.yaml?

[itsmano1993]

which version of feast are you using?

feast==0.43.0

even I installed latest version of feast 0.45.0 and still experiencing the same issue

[itsmano1993]

can you share your feature_store.yaml?

---
project: dev
registry: data/registry.db
provider: local
online_store:
  type: mysql
  host: localhost
  port: 3306
  database: feast
  user: progressfinQA
  password: passwordQA
entity_key_serialization_version: 2
auth:
  type: no_auth

[itsmano1993]

@franciscojavierarceo Did you get a chance to look into this?