feature_retrieval

Expand source code

import sys
from datetime import datetime

import feast
import pandas as pd


def retrieve_historical_feature_demo(repo_path: str):
    income_entities = pd.DataFrame.from_dict(
        {
            "ifa": [
                "27a",
                "30a",
                "475a",
                "965a",
                "1678a",
                "1698a",
                "1807a",
                "1951a",
                "2041a",
                "2215a",
            ],
            "event_time": [
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
            ],
        }
    )

    fs = feast.FeatureStore(repo_path=repo_path)
    income_features_df = fs.get_historical_features(
        entity_df=income_entities,
        features=[
            "income_view:income",
            "income_view:latent_0",
            "income_view:latent_1",
            "income_view:latent_2",
            "income_view:latent_3",
        ],
    ).to_df()
    print(income_features_df.head())

    # train model from here ...

    feature_service = fs.get_feature_service("income_feature_service")
    income_features_by_service_df = fs.get_historical_features(
        features=feature_service, entity_df=income_entities
    ).to_df()
    print(income_features_by_service_df.head())


if __name__ == "__main__":
    if len(sys.argv) < 2:
        print("Please, provide a path to anovos feature repo!")
        exit(1)
    path = sys.argv[1]
    retrieve_historical_feature_demo(repo_path=path)

Functions

def retrieve_historical_feature_demo(repo_path: str)

Expand source code

def retrieve_historical_feature_demo(repo_path: str):
    income_entities = pd.DataFrame.from_dict(
        {
            "ifa": [
                "27a",
                "30a",
                "475a",
                "965a",
                "1678a",
                "1698a",
                "1807a",
                "1951a",
                "2041a",
                "2215a",
            ],
            "event_time": [
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
                datetime.now(),
            ],
        }
    )

    fs = feast.FeatureStore(repo_path=repo_path)
    income_features_df = fs.get_historical_features(
        entity_df=income_entities,
        features=[
            "income_view:income",
            "income_view:latent_0",
            "income_view:latent_1",
            "income_view:latent_2",
            "income_view:latent_3",
        ],
    ).to_df()
    print(income_features_df.head())

    # train model from here ...

    feature_service = fs.get_feature_service("income_feature_service")
    income_features_by_service_df = fs.get_historical_features(
        features=feature_service, entity_df=income_entities
    ).to_df()
    print(income_features_by_service_df.head())