On-Demand Feature Usage¶

In [0]:

import tecton
import pandas
from datetime import datetime

ws = tecton.get_workspace("prod")

import tecton import pandas from datetime import datetime ws = tecton.get_workspace("prod")

On-Demand Feature Views without Dependencies¶

Load an On-Demand Feature View¶

In [0]:

fv = ws.get_feature_view("transaction_amount_is_high")
fv.summary()

fv = ws.get_feature_view("transaction_amount_is_high") fv.summary()

Out[3]:

Name	transaction_amount_is_high
Workspace	prod
Description	The transaction amount is higher than $100.
Created At	2022-06-06 23:13:49 UTC
Owner
Last Modified By	jake@tecton.ai
Family
Source Filename	fraud/features/on_demand_feature_views/transaction_amount_is_high.py
Tags	{}
Type	OnDemandFeatureView
URL	https://staging.tecton.ai/app/repo/prod/features/transaction_amount_is_high
Entities
Features	transaction_amount_is_high
Transformation	transaction_amount_is_high
Join Keys
Request Context Keys	amt

Execute an On-Demand Feature View Online¶

Because this On-Demand Feature View has no dependencies it only needs request data to execute

In [0]:

fv.get_online_features(request_data={"amt": 17000}).to_dict()

fv.get_online_features(request_data={"amt": 17000}).to_dict()

Out[8]: {'transaction_amount_is_high': True}

Execute On-Demand Feature View Offline¶

In [0]:

spine_df = pandas.DataFrame({'amt': [200, 20]})
features_df = fv.get_historical_features(spine=spine_df).to_pandas()
display(features_df)

spine_df = pandas.DataFrame({'amt': [200, 20]}) features_df = fv.get_historical_features(spine=spine_df).to_pandas() display(features_df)

INFO - 06/07/2022 06:00:58 PM - Query Builder - transaction_amount_is_high: (FV 1/1) Start Build [1/3] INFO - 06/07/2022 06:00:58 PM - Query Builder - transaction_amount_is_high: Building Tile DataFrame [2/3]

amt	transaction_amount_is_high__transaction_amount_is_high
200	true
20	false

On-Demand Feature Views with Feature View Dependencies¶

Load an On-Demand Feature View¶

In [0]:

fv = ws.get_feature_view("transaction_amount_is_higher_than_average")

fv = ws.get_feature_view("transaction_amount_is_higher_than_average")

Run Feature View Transformation Pipeline with Mock Inputs¶

To use run on On-Demand Feature View, all Feature View inputs must be mocked.

In [0]:

transaction_request = pandas.DataFrame([{
  "amt": 100.0
}])

user_transaction_amount_metrics = pandas.DataFrame([{
  "amt_mean_1d_10m": 200.0
}])

result = fv.run(transaction_request=transaction_request, user_transaction_amount_metrics=user_transaction_amount_metrics)
print(result)

transaction_request = pandas.DataFrame([{ "amt": 100.0 }]) user_transaction_amount_metrics = pandas.DataFrame([{ "amt_mean_1d_10m": 200.0 }]) result = fv.run(transaction_request=transaction_request, user_transaction_amount_metrics=user_transaction_amount_metrics) print(result)

{'transaction_amount_is_higher_than_average': 0 False dtype: bool}

Execute On-Demand FeatureView with Dependencies Offline¶

Because this On-Demand Feature View depends on another Feature View, we will first preview the dependent Feature View. We can use this to select keys that will be needed for historical lookups in order for the On-Demand Feature View to run. The dependent column in the Feature View below is amt_mean_1d_10m.

In [0]:

dependent_fv = ws.get_feature_view("user_transaction_amount_metrics")
dependent_fv.get_historical_features(start_time=datetime(2022, 5, 1), end_time=datetime(2022, 5, 2)).to_pandas().sample(4)

dependent_fv = ws.get_feature_view("user_transaction_amount_metrics") dependent_fv.get_historical_features(start_time=datetime(2022, 5, 1), end_time=datetime(2022, 5, 2)).to_pandas().sample(4)

Out[21]:

	user_id	amt_sum_1h_10m	amt_sum_1d_10m	amt_sum_3d_10m	amt_mean_1h_10m	amt_mean_1d_10m	amt_mean_3d_10m	timestamp
4350	user_724235628997	1866.68	82881.33	231085.28	46.667000	71.820910	64.548961	2022-05-01 08:20:00
3452	user_650387977076	3803.60	123012.79	360482.51	63.393333	80.982745	78.331706	2022-05-01 20:10:00
5261	user_916905857181	857.92	31343.50	96362.05	71.493333	95.559451	96.073829	2022-05-01 10:20:00
483	user_222506789984	7054.43	140996.11	424768.41	69.845842	66.570401	67.713759	2022-05-01 19:00:00

In [0]:

spine_df = pandas.DataFrame({
  'user_id': ["user_724235628997", "user_650387977076", "user_916905857181", "user_222506789984"],
  'timestamp': [datetime(2022, 5, 1, 8, 20), datetime(2022, 5, 1, 20, 10), datetime(2022, 5, 1, 10, 20), datetime(2022, 5, 1, 19, 0),],
  'amt': [(71.820910 + 10), (80.982745 - 10), (95.559451 + 10), (66.570401 - 10)]
})
spine_df.head()

spine_df = pandas.DataFrame({ 'user_id': ["user_724235628997", "user_650387977076", "user_916905857181", "user_222506789984"], 'timestamp': [datetime(2022, 5, 1, 8, 20), datetime(2022, 5, 1, 20, 10), datetime(2022, 5, 1, 10, 20), datetime(2022, 5, 1, 19, 0),], 'amt': [(71.820910 + 10), (80.982745 - 10), (95.559451 + 10), (66.570401 - 10)] }) spine_df.head()

Out[24]:

	user_id	timestamp	amt
0	user_724235628997	2022-05-01 08:20:00	81.820910
1	user_650387977076	2022-05-01 20:10:00	70.982745
2	user_916905857181	2022-05-01 10:20:00	105.559451
3	user_222506789984	2022-05-01 19:00:00	56.570401

In [0]:

fv.get_historical_features(spine=spine_df).to_pandas()

fv.get_historical_features(spine=spine_df).to_pandas()

INFO - 06/07/2022 06:11:55 PM - Query Builder - user_transaction_amount_metrics: (FV 1/2) Start Build [1/3] INFO - 06/07/2022 06:11:55 PM - Query Builder - user_transaction_amount_metrics: Building Tile DataFrame [2/3] INFO - 06/07/2022 06:12:13 PM - Query Builder - user_transaction_amount_metrics: Generating Join [3/3] INFO - 06/07/2022 06:12:13 PM - Query Builder - transaction_amount_is_higher_than_average: (FV 2/2) Start Build [1/3] INFO - 06/07/2022 06:12:13 PM - Query Builder - transaction_amount_is_higher_than_average: Building Tile DataFrame [2/3] Out[25]:

	user_id	timestamp	amt	transaction_amount_is_higher_than_average__transaction_amount_is_higher_than_average
0	user_724235628997	2022-05-01 08:20:00	81.820910	True
1	user_650387977076	2022-05-01 20:10:00	70.982745	False
2	user_916905857181	2022-05-01 10:20:00	105.559451	True
3	user_222506789984	2022-05-01 19:00:00	56.570401	False

Execute On-Demand FeatureView with Dependencies Online¶

In [0]:

# This will compare a transaction amount against the user's most recent average 24h transaction amount.
fv.get_online_features(request_data={"amt": 150}, join_keys={"user_id": "user_724235628997"}).to_dict()

# This will compare a transaction amount against the user's most recent average 24h transaction amount. fv.get_online_features(request_data={"amt": 150}, join_keys={"user_id": "user_724235628997"}).to_dict()

Out[26]: {'transaction_amount_is_higher_than_average': True}