4. Update Descriptions to Tables

(Optional) Updating descriptions to tables and columns¶

Table and column descriptions are automatically fetched from your Data Warehouse when they are available. If these descriptions are missing or incomplete, you have the option to edit and update them.

While not mandatory, updating concise descriptions to tables and columns can be immensely beneficial if you are using FeatureByte Enterprise. These annotations assist FeatureByte's Feature Ideation engine in generating insightful features.

Much like a data scientist, FeatureByte does its best to grasp the significance and purpose of various tables and columns, discerning their types and more. Based on this understanding, it suggests pertinent aggregations and feature combinations. While FeatureByte can operate effectively without these descriptions, having them certainly enhances the quality of its recommendations.

In [1]:

import featurebyte as fb

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

catalog_name = "Credit Default 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 = "Credit Default Dataset SDK Tutorial" catalog = fb.Catalog.activate(catalog_name)

16:38:24 | WARNING  | Service endpoint is inaccessible: http://featurebyte-server:8088/
16:38:24 | INFO     | Using profile: tutorial
16:38:24 | INFO     | Using configuration file at: /Users/gxav/.featurebyte/config.yaml
16:38:24 | INFO     | Active profile: tutorial (https://tutorials.featurebyte.com/api/v1)
16:38:24 | INFO     | SDK version: 2.1.0.dev113
16:38:24 | INFO     | No catalog activated.
16:38:24 | INFO     | Catalog activated: Credit Default Dataset SDK Tutorial
16:05:56 | WARNING  | Remote SDK version (1.1.0.dev7) is different from local (1.1.0.dev1). Update local SDK to avoid unexpected behavior.
16:05:56 | INFO     | No catalog activated.
16:05:56 | INFO     | Catalog activated: Grocery Dataset Tutorial

Get tables from the catalog first¶

In [2]:

catalog.list_tables()

catalog.list_tables()

Out[2]:

	id	name	type	status	entities	created_at
0	67c2c752924afe7a79ec6f27	CONSUMER_INSTALLMENTS	time_series_table	PUBLIC_DRAFT	[Consumer Loan]	2025-03-01T08:37:38.763000
1	67c2c750924afe7a79ec6f26	CONSUMER_LOAN_STATUS	scd_table	PUBLIC_DRAFT	[Consumer Loan, Client]	2025-03-01T08:37:36.903000
2	67c2c74e924afe7a79ec6f25	PRIOR_APPLICATIONS	event_table	PUBLIC_DRAFT	[Prior Application, Client]	2025-03-01T08:37:34.829000
3	67c2c74c924afe7a79ec6f24	NEW_APPLICATION	dimension_table	PUBLIC_DRAFT	[New Application, Client]	2025-03-01T08:37:32.806000

In [3]:

new_application = catalog.get_table("NEW_APPLICATION")
prior_applications = catalog.get_table("PRIOR_APPLICATIONS")
consumer_loans = catalog.get_table("CONSUMER_LOAN_STATUS")
consumer_installments = catalog.get_table("CONSUMER_INSTALLMENTS")

new_application = catalog.get_table("NEW_APPLICATION") prior_applications = catalog.get_table("PRIOR_APPLICATIONS") consumer_loans = catalog.get_table("CONSUMER_LOAN_STATUS") consumer_installments = catalog.get_table("CONSUMER_INSTALLMENTS")

Discover the current descriptions of tables¶

In [4]:

new_application.description

new_application.description

Out[4]:

'Records new loan applications.'

In [5]:

prior_applications.description

prior_applications.description

Out[5]:

'Contains data on prior loan applications and the final decision.'

In [6]:

consumer_loans.description

consumer_loans.description

Out[6]:

'Tracks consumer loans status.'

In [7]:

consumer_installments.description

consumer_installments.description

Out[7]:

'Logs monthly installments for consumer loans at the time of payment.'

Update descriptions of one table¶

In [8]:

new_application.update_description('new loan applications.')

new_application.update_description('new loan applications.')

In [9]:

new_application.description

new_application.description

Out[9]:

'new loan applications.'

In [10]:

new_application.update_description('application')

new_application.update_description('application')

Discover the current descriptions of columns for NEW_APPLICATION¶

You can either display all columns together

In [11]:

import pandas as pd
pd.DataFrame(new_application.info(verbose=True)["columns_info"])

import pandas as pd pd.DataFrame(new_application.info(verbose=True)["columns_info"])

Out[11]:

	name	dtype	entity	semantic	critical_data_info	description
0	BIRTHDATE	VARCHAR	None	None	None	Client birthdate
1	CODE_GENDER	VARCHAR	None	None	None	Gender of the client
2	INCOME_TYPE	VARCHAR	None	None	None	Clients income type (businessman, working, mat...
3	EDUCATION_TYPE	VARCHAR	None	None	None	Level of highest education the client achieved
4	OCCUPATION_TYPE	VARCHAR	None	None	None	What kind of occupation does the client have
5	ORGANIZATION_TYPE	VARCHAR	None	None	None	Type of organization where client works
6	APPLICATION_ID	INT	New Application	dimension_id	None	ID of application
7	CLIENT_ID	INT	Client	None	None	ID of the client
8	AMT_CREDIT	FLOAT	None	None	None	Credit amount of the loan
9	AMT_ANNUITY	FLOAT	None	None	None	Loan annuity
10	AMT_GOODS_VALUE	FLOAT	None	None	None	For consumer loans it is the value of the good...
11	REGION_POPULATION_RELATIVE	FLOAT	None	None	None	Normalized population of region where client l...
12	APPLICATION_TIME	TIMESTAMP	None	None	None	Application timestamp
13	DAYS_EMPLOYED	INT	None	None	None	How many days before the application the perso...
14	DAYS_REGISTRATION	FLOAT	None	None	None	How many days before the application did clien...
15	DAYS_LAST_PHONE_CHANGE	FLOAT	None	None	None	How many days before the application did clien...
16	FLOORSMAX_MEDI	FLOAT	None	None	None	Normalized information about building where th...
17	LANDAREA_MEDI	FLOAT	None	None	None	Normalized information about building where th...
18	FLAG_DOCUMENT_3	INT	None	None	None	Did client provide document
19	AMT_REQ_CREDIT_BUREAU_QRT	FLOAT	None	None	None	Number of enquiries to Credit Bureau about the...
20	available_at	TIMESTAMP	None	record_creation_timestamp	None	Timestamp the record was added to the database

Or display each column one by one

In [12]:

for column in new_application.columns:
    print(f"{column}: {new_application[column].description}")

for column in new_application.columns: print(f"{column}: {new_application[column].description}")

BIRTHDATE: Client birthdate
CODE_GENDER: Gender of the client
INCOME_TYPE: Clients income type (businessman, working, maternity leave,…)
EDUCATION_TYPE: Level of highest education the client achieved
OCCUPATION_TYPE: What kind of occupation does the client have
ORGANIZATION_TYPE: Type of organization where client works
APPLICATION_ID: ID of application
CLIENT_ID: ID of the client
AMT_CREDIT: Credit amount of the loan
AMT_ANNUITY: Loan annuity
AMT_GOODS_VALUE: For consumer loans it is the value of the goods for which the loan is given
REGION_POPULATION_RELATIVE: Normalized population of region where client lives (higher number means the client lives in more populated region)
APPLICATION_TIME: Application timestamp
DAYS_EMPLOYED: How many days before the application the person started current employment
DAYS_REGISTRATION: How many days before the application did client change his registration
DAYS_LAST_PHONE_CHANGE: How many days before the application did client change his phone
FLOORSMAX_MEDI: Normalized information about building where the client lives
LANDAREA_MEDI: Normalized information about building where the client lives
FLAG_DOCUMENT_3: Did client provide document
AMT_REQ_CREDIT_BUREAU_QRT: Number of enquiries to Credit Bureau about the client 3 month before application (excluding one month before application)
available_at: Timestamp the record was added to the database

Update column descriptions¶

If the description is incorrect or incomplete, you can edit it

In [13]:

# By using the table method: update_column_description
new_application.update_column_description(
    "DAYS_EMPLOYED",
    "How many days before the application the person started current employment"
)

# By using the table method: update_column_description new_application.update_column_description( "DAYS_EMPLOYED", "How many days before the application the person started current employment" )

In [14]:

# Or by using the column method: update_description
new_application.DAYS_EMPLOYED.update_description(
    "How many days before the application the person started current employment"
)

# Or by using the column method: update_description new_application.DAYS_EMPLOYED.update_description( "How many days before the application the person started current employment" )

That's it for this tutorial. Again, this is an optional step, but it can drastically improve FeatureByte's feature ideation.

SDK reference for¶

• Table.info