Getting Features for Training
In Getting Started, we showed you how to retrieve a training data sample from the Feature Store.
Below, we show this example again:
from qwak.feature_store.offline import OfflineFeatureStore
offline_feature_store = OfflineFeatureStore()
data = offline_feature_store.get_sample_data(feature_set_name='user_credit_risk_features_v2', number_of_rows=999)
Alternatively, we can retrieve data from the OfflineFeatureStore
by entity id and the last modification timestamp.
In this case, we must define:
- The filter DataFrame containing the entity id and the point-in-time column name.
- A list of features that we want to retrieve.
import pandas as pd
from qwak.feature_store.offline import OfflineFeatureStore
df = pd.DataFrame(columns=[ 'user_id', 'timestamp' ],
data =[[ '06cc255a-aa07-4ec9-ac69-b896ccf05322', '2021-01-01 00:00:00']])
key_to_features = {'user_id': ['user_credit_risk_features_v2.checking_account',
'user_credit_risk_features_v2.age',
'user_credit_risk_features_v2.job']}
offline_feature_store = OfflineFeatureStore()
train_df = offline_feature_store.get_feature_values(
entity_key_to_features=key_to_features,
population=df,
point_in_time_column_name='timestamp')
Updated 10 months ago