from feast import FeatureStore, RepoConfig

from feast.repo_config import RegistryConfig

# TODO: replace with your bucket

repo_config = RepoConfig(

registry=RegistryConfig(path="gs://feast-workshop-danny/registry.pb"),

project="feast_demo_gcp",

provider="gcp",

offline_store="file", # Could also be the OfflineStoreConfig e.g. FileOfflineStoreConfig

online_store="null", # Could also be the OnlineStoreConfig e.g. RedisOnlineStoreConfig

)

store = FeatureStore(config=repo_config)

import pandas as pd

# Get the latest feature values for unique entities

entity_df = pd.DataFrame.from_dict({"driver_id": [1001, 1002, 1003, 1004, 1005],})

entity_df["event_timestamp"] = pd.to_datetime("now", utc=True)

training_df = store.get_historical_features(

entity_df=entity_df, features=store.get_feature_service("model_v2"),

).to_df()

# Make batch predictions

# predictions = model.predict(training_df)

print(training_df)