Skip to content

3. Register Entities

What is an Entity?

In FeatureByte, an Entity represents real-world objects or concepts. While entities are typically associated with a primary key or foreign key in the data, they can also be represented by categorical columns that define groups of related objects. For example, a City entity may represent multiple customers, and a Product Group entity may encompass multiple products, even though neither is explicitly used as a foreign key.

Step 1: Identify Entities in Your Data

In the Credit Default dataset, we identified six entities.

Name Description
New Application A new loan application
Client A client of the credit institution
BureauReportedCredit A report from the Credit Bureau
PriorApplication A prior loan application
Loan A loan
Installment An installment payment

Step 2: Decide their Serving Name

An Entity's Serving Name is the unique identifier used to recognize the entity during a serving request in FeatureByte. You can choose any name that suits your needs, provided it is not already assigned to another entity.

Name Description Serving Name
New Application A new loan application SK_ID_CURR
Client A client of the credit institution ClientID
BureauReportedCredit A report from the Credit Bureau SK_ID_BUREAU
PriorApplication A prior loan application APPLICATION_ID
Loan A loan LOAN_ID
Installment An installment payment INSTALMENT_ID

Step 3: Register Entities

Navigate to the Entity catalog from the 'Explore' section of the menu.

Repeat this for each entity:

  1. Click Register Entities.
  2. Fill in the form: the Entity's Name, its Serving Name and optionally its description.
  3. Confirm by clicking Save Entity.

Register Entities

Register Entities

Register Entities

Register Entities

Register Entities

Register Entities

Step 4: Review Registered Entities

Verify the registration by checking the Entity Catalog under the 'Explore' section.

Register Entities

Step 5: Tag Columns Representing Entities

From the menu, navigate to the 'Explore' section and access the Table catalog.

Table Description

For each table, navigate to the 'Columns' tab and associate relevant columns with the entities they represent.

Tag Entity

In NEW_APPLICATION,

  • SK_ID_CURR represents New Application.
  • ClientID identifies Client.

Register Entities In Table

In CLIENT_PROFILE,

  • ClientID identifies Client.

Register Entities In Table

In BUREAU,

  • SK_ID_BUREAU represents BureauReportedCredit.
  • ClientID identifies Client.

Register Entities In Table

In PREVIOUS_APPLICATIONS,

  • APPLICATION_ID represents PriorApplication.
  • ClientID identifies Client.

Register Entities In Table

In LOAN_STATUS,

  • LOAN_ID represents Loan.
  • APPLICATION_ID identifies PriorApplication.

Register Entities In Table

In INSTALLMENTS_PAYMENTS,

  • INSTALMENT_ID represents Installment.
  • APPLICATION_ID represents PriorApplication.

Register Entities In Table

In CREDIT_CARD_MONTHLY_BALANCE,

  • ClientID represents Client.

Register Entities In Table

Step 6: Review Entities Tagging and Relationships

  1. Review entity tagging in the Entity Catalog.

    Entity Calalog

  2. Verify relationships in the Relationships Catalog.

    Open the Relationships Catalog to ensure that all relationships created through tagging are correct.

    If multiple tables could define the relationship between two entities, update the Relation Table to specify the table that should serve as the authoritative source.

    Relationships

  3. Validate the table diagram in the Table Catalog.

    Navigate to the Table Catalog.

    Table Catalog

    Click the diagram icon in Table Diagramto review the automatically generated table diagram.

Table Diagram