3. Register Entities

Registering entities represented in Loan Applications dataset¶

In FeatureByte, an entity models real-world objects and ideas. These entities often correspond to columns in database tables.

Taking our loan applications scenario as an example, we can view "Client", "New Application", "Prior Application", and "Loan" as entities.

To help FeatureByte identify these entities in the data and the columns that represent them, we'll be creating and tagging these entities in this tutorial.

In [1]:

            
                Copied!
                
import featurebyte as fb

# Set your profile to the tutorial environment
fb.use_profile("tutorial")

catalog_name = "Loan Applications Dataset SDK Tutorial"
catalog = fb.Catalog.activate(catalog_name)
import featurebyte as fb

# Set your profile to the tutorial environment
fb.use_profile("tutorial")

catalog_name = "Loan Applications Dataset SDK Tutorial"
catalog = fb.Catalog.activate(catalog_name)

14:06:37 | INFO     | SDK version: 3.0.1.dev45
INFO    :featurebyte:SDK version: 3.0.1.dev45
14:06:37 | INFO     | No catalog activated.
INFO    :featurebyte:No catalog activated.
14:06:37 | INFO     | Using profile: tutorial
INFO    :featurebyte:Using profile: tutorial
14:06:37 | INFO     | Using configuration file at: /Users/gxav/.featurebyte/config.yaml
INFO    :featurebyte:Using configuration file at: /Users/gxav/.featurebyte/config.yaml
14:06:37 | INFO     | Active profile: tutorial (https://tutorials.featurebyte.com/api/v1)
INFO    :featurebyte:Active profile: tutorial (https://tutorials.featurebyte.com/api/v1)
14:06:37 | INFO     | SDK version: 3.0.1.dev45
INFO    :featurebyte:SDK version: 3.0.1.dev45
14:06:37 | INFO     | No catalog activated.
INFO    :featurebyte:No catalog activated.
14:06:37 | INFO     | Catalog activated: Loan Applications Dataset SDK Tutorial
INFO    :featurebyte.api.catalog:Catalog activated: Loan Applications Dataset SDK Tutorial
16:05:47 | WARNING  | Remote SDK version (1.1.0.dev7) is different from local (1.1.0.dev1). Update local SDK to avoid unexpected behavior.
16:05:47 | INFO     | No catalog activated.
16:05:47 | INFO     | Catalog activated: Grocery Dataset Tutorial

Identify Entities in Your Data and Decide their Serving Name¶

When creating an entity, you'll need to define its serving name. This name acts as a unique identifier, particularly during preview or serving requests.

In [2]:

            
                Copied!
                
                    
                    
                
                

        
catalog.create_entity(name="New Application", serving_names=["SK_ID_CURR"])
catalog.create_entity(name="Client", serving_names=["ClientID"])
catalog.create_entity(name="BureauReportedCredit", serving_names=["SK_ID_BUREAU"])
catalog.create_entity(name="PriorApplication", serving_names=["APPLICATION_ID"])
catalog.create_entity(name="Installment", serving_names=["INSTALMENT_ID"])
catalog.create_entity(name="Loan", serving_names=["LOAN_ID"])
catalog.create_entity(name="New Application", serving_names=["SK_ID_CURR"])
catalog.create_entity(name="Client", serving_names=["ClientID"])
catalog.create_entity(name="BureauReportedCredit", serving_names=["SK_ID_BUREAU"])
catalog.create_entity(name="PriorApplication", serving_names=["APPLICATION_ID"])
catalog.create_entity(name="Installment", serving_names=["INSTALMENT_ID"])
catalog.create_entity(name="Loan", serving_names=["LOAN_ID"])

Out[2]:

Entity

name	Loan
created_at	2025-06-02 05:47:03
updated_at	None
description	None
serving_names	['LOAN_ID']
catalog_name	Loan Applications Dataset SDK Tutorial

Tag Columns Representing Entities¶

Now that we've established the entities, it's time to guide FeatureByte in mapping these entities to the actual data in our tables.

In [3]:

            
                Copied!
                
                    
                    
                
                

        
application = catalog.get_table("NEW_APPLICATION")
application["ClientID"].as_entity("Client")
application["SK_ID_CURR"].as_entity("New Application")

client_profile = catalog.get_table("CLIENT_PROFILE")
client_profile["ClientID"].as_entity("Client")

bureau = catalog.get_table("BUREAU")
bureau["ClientID"].as_entity("Client")
bureau["SK_ID_BUREAU"].as_entity("BureauReportedCredit")

previous_application = catalog.get_table("PREVIOUS_APPLICATION")
previous_application["ClientID"].as_entity("Client")
previous_application["APPLICATION_ID"].as_entity("PriorApplication")

loan_status = catalog.get_table("LOAN_STATUS")
loan_status["LOAN_ID"].as_entity("Loan")
loan_status['APPLICATION_ID'].as_entity("PriorApplication")

installments_payments = catalog.get_table("INSTALLMENTS_PAYMENTS")
installments_payments["INSTALMENT_ID"].as_entity("Installment")
installments_payments["APPLICATION_ID"].as_entity("PriorApplication")

credit_card_balance = catalog.get_table("CREDIT_CARD_MONTHLY_BALANCE")
credit_card_balance["ClientID"].as_entity("Client")
application = catalog.get_table("NEW_APPLICATION")
application["ClientID"].as_entity("Client")
application["SK_ID_CURR"].as_entity("New Application")

client_profile = catalog.get_table("CLIENT_PROFILE")
client_profile["ClientID"].as_entity("Client")

bureau = catalog.get_table("BUREAU")
bureau["ClientID"].as_entity("Client")
bureau["SK_ID_BUREAU"].as_entity("BureauReportedCredit")

previous_application = catalog.get_table("PREVIOUS_APPLICATION")
previous_application["ClientID"].as_entity("Client")
previous_application["APPLICATION_ID"].as_entity("PriorApplication")

loan_status = catalog.get_table("LOAN_STATUS")
loan_status["LOAN_ID"].as_entity("Loan")
loan_status['APPLICATION_ID'].as_entity("PriorApplication")

installments_payments = catalog.get_table("INSTALLMENTS_PAYMENTS")
installments_payments["INSTALMENT_ID"].as_entity("Installment")
installments_payments["APPLICATION_ID"].as_entity("PriorApplication")

credit_card_balance = catalog.get_table("CREDIT_CARD_MONTHLY_BALANCE")
credit_card_balance["ClientID"].as_entity("Client")

Review Entities Relationships¶

Now, if we list the tables as we did in the previous tutorial, we'll notice that entities have been assigned to each table.

In [4]:

            
                Copied!
                
display(catalog.list_tables())
display(catalog.list_tables())

	id	name	type	status	entities	created_at
0	683d3a70f4d55cd61d526cee	CREDIT_CARD_MONTHLY_BALANCE	time_series_table	PUBLIC_DRAFT	[Client]	2025-06-02T05:45:20.813000
1	683d3a6ef4d55cd61d526ced	LOAN_STATUS	scd_table	PUBLIC_DRAFT	[Loan, PriorApplication]	2025-06-02T05:45:18.744000
2	683d3a6cf4d55cd61d526cec	PREVIOUS_APPLICATION	event_table	PUBLIC_DRAFT	[PriorApplication, Client]	2025-06-02T05:45:16.825000
3	683d3a6af4d55cd61d526ceb	INSTALLMENTS_PAYMENTS	event_table	PUBLIC_DRAFT	[Installment, PriorApplication]	2025-06-02T05:45:14.598000
4	683d3a69f4d55cd61d526cea	BUREAU	event_table	PUBLIC_DRAFT	[Client, BureauReportedCredit]	2025-06-02T05:45:13.206000
5	683d3a66f4d55cd61d526ce9	CLIENT_PROFILE	scd_table	PUBLIC_DRAFT	[Client]	2025-06-02T05:45:11.014000
6	683d3a64f4d55cd61d526ce8	NEW_APPLICATION	dimension_table	PUBLIC_DRAFT	[New Application, Client]	2025-06-02T05:45:08.853000

We can also list entities separately:

In [5]:

            
                Copied!
                
display(catalog.list_entities())
display(catalog.list_entities())

	id	name	serving_names	created_at
0	683d3ad761061d4da44450d5	Loan	[LOAN_ID]	2025-06-02T05:47:03.194000
1	683d3ad661061d4da44450d4	Installment	[INSTALMENT_ID]	2025-06-02T05:47:03.049000
2	683d3ad661061d4da44450d3	PriorApplication	[APPLICATION_ID]	2025-06-02T05:47:02.918000
3	683d3ad661061d4da44450d2	BureauReportedCredit	[SK_ID_BUREAU]	2025-06-02T05:47:02.783000
4	683d3ad661061d4da44450d1	Client	[ClientID]	2025-06-02T05:47:02.638000
5	683d3ad661061d4da44450d0	New Application	[SK_ID_CURR]	2025-06-02T05:47:02.488000

And let's examine the relationships between entities, which FeatureByte has conveniently outlined for us:

In [6]:

            
                Copied!
                
display(catalog.list_relationships())
display(catalog.list_relationships())

	id	relationship_type	entity	related_entity	relation_table	relation_table_type	enabled	created_at	updated_at
0	683d3ada954f0aa89942ada0	child_parent	Installment	PriorApplication	INSTALLMENTS_PAYMENTS	event_table	True	2025-06-02T05:47:06.230000	None
1	683d3ad9954f0aa89942ad98	child_parent	Loan	PriorApplication	LOAN_STATUS	scd_table	True	2025-06-02T05:47:05.761000	None
2	683d3ad9954f0aa89942ad8f	child_parent	PriorApplication	Client	PREVIOUS_APPLICATION	event_table	True	2025-06-02T05:47:05.245000	None
3	683d3ad8954f0aa89942ad89	child_parent	BureauReportedCredit	Client	BUREAU	event_table	True	2025-06-02T05:47:04.739000	None
4	683d3ad7cd54c1e3f75b0dfc	child_parent	New Application	Client	NEW_APPLICATION	dimension_table	True	2025-06-02T05:47:03.867000	None

That's it for most important data modeling concepts in FeatureByte. In next tutorials we will look into concepts related to data cleaning and feature engineering.

For more details about concepts we discussed in this tutorial, feel free to explore:¶

More info on Entities