PRODUCT vs OVERALL item TotalCost across customer Age bands 28d

SDK code to create PRODUCT_vs_OVERALL_item_TotalCost_across_customer_Age_bands_28d¶

Feature description:

Similarity between the product and all products measured by the Cosine Similarity between the Distribution representing the cumulative TotalCost of item, categorized by their respective customer_Age_band, over 28d for both entities.

In [ ]:

import featurebyte as fb
fb.use_profile("tutorial")

import featurebyte as fb fb.use_profile("tutorial")

Activate catalog¶

See SDK reference for catalogs

In [ ]:

catalog = fb.Catalog.activate("Grocery Dataset Tutorial")

catalog = fb.Catalog.activate("Grocery Dataset Tutorial")

Set windows for aggregation¶

In [ ]:

windows = ['28d']

windows = ['28d']

Get view from table¶

See SDK reference for views

In [ ]:

# Get view from INVOICEITEMS item table.
invoiceitems_view = catalog.get_view("INVOICEITEMS")

# Get view from INVOICEITEMS item table. invoiceitems_view = catalog.get_view("INVOICEITEMS")

In [ ]:

# Get view from GROCERYCUSTOMER scd table.
grocerycustomer_view = catalog.get_view("GROCERYCUSTOMER")

# Get view from GROCERYCUSTOMER scd table. grocerycustomer_view = catalog.get_view("GROCERYCUSTOMER")

Join views¶

See SDK reference for views

In [ ]:

# Join GROCERYCUSTOMER view to INVOICEITEMS view.
invoiceitems_view = invoiceitems_view.join(grocerycustomer_view, rsuffix="")

# Join GROCERYCUSTOMER view to INVOICEITEMS view. invoiceitems_view = invoiceitems_view.join(grocerycustomer_view, rsuffix="")

Derive Timedelta from datetime column¶

See SDK reference for view columns

In [ ]:

# Derive Age from the event timestamp and the date of birth.
invoiceitems_view["customer_Age"] = (
    (
        invoiceitems_view["Timestamp"]
        - invoiceitems_view["DateOfBirth"]
    ).dt.day / 365.25
).floor()
# Transform age into a 5 year age band.
invoiceitems_view["customer_Age_band"] = (
    ((invoiceitems_view["customer_Age"] + 1) / 5).ceil() - 1
) * 5
invoiceitems_view["customer_Age_band"] = (
    invoiceitems_view["customer_Age_band"].astype(str)
     + "-" + (invoiceitems_view["customer_Age_band"] + 4).astype(str)
)

# Derive Age from the event timestamp and the date of birth. invoiceitems_view["customer_Age"] = ( ( invoiceitems_view["Timestamp"] - invoiceitems_view["DateOfBirth"] ).dt.day / 365.25 ).floor() # Transform age into a 5 year age band. invoiceitems_view["customer_Age_band"] = ( ((invoiceitems_view["customer_Age"] + 1) / 5).ceil() - 1 ) * 5 invoiceitems_view["customer_Age_band"] = ( invoiceitems_view["customer_Age_band"].astype(str) + "-" + (invoiceitems_view["customer_Age_band"] + 4).astype(str) )

Do window aggregation from INVOICEITEMS¶

See SDK reference for features

See SDK reference to groupby a view

See SDK reference to do aggregation over time

See SDK reference for feature group

See list of aggregation operations

In [ ]:

# Group INVOICEITEMS view by product entity (GroceryProductGuid) across different
# customer_Age_bands.
invoiceitems_view_by_product_across_customer_age_band =\
invoiceitems_view.groupby(
    ['GroceryProductGuid'], category="customer_Age_band"
)

# Group INVOICEITEMS view by product entity (GroceryProductGuid) across different # customer_Age_bands. invoiceitems_view_by_product_across_customer_age_band =\ invoiceitems_view.groupby( ['GroceryProductGuid'], category="customer_Age_band" )

In [ ]:

# Distribution representing the cumulative TotalCost of item, categorized by their respective
# customer_Age_band, for the product over time.
feature_group =\
invoiceitems_view_by_product_across_customer_age_band.aggregate_over(
    "TotalCost", method=fb.AggFunc.SUM,
    feature_names=[
        "PRODUCT_item_TotalCost_across_customer_Age_bands"
        + "_" + w for w in windows
    ],
    windows=windows
)
# Get PRODUCT_item_TotalCost_across_customer_Age_bands_28d object from feature group.
product_item_totalcost_across_customer_age_bands_28d =\
feature_group["PRODUCT_item_TotalCost_across_customer_Age_bands_28d"]

# Distribution representing the cumulative TotalCost of item, categorized by their respective # customer_Age_band, for the product over time. feature_group =\ invoiceitems_view_by_product_across_customer_age_band.aggregate_over( "TotalCost", method=fb.AggFunc.SUM, feature_names=[ "PRODUCT_item_TotalCost_across_customer_Age_bands" + "_" + w for w in windows ], windows=windows ) # Get PRODUCT_item_TotalCost_across_customer_Age_bands_28d object from feature group. product_item_totalcost_across_customer_age_bands_28d =\ feature_group["PRODUCT_item_TotalCost_across_customer_Age_bands_28d"]

In [ ]:

# Group INVOICEITEMS view across different customer_Age_bands.
invoiceitems_view_by_overall_across_customer_age_band =\
invoiceitems_view.groupby(
    [], category="customer_Age_band"
)

# Group INVOICEITEMS view across different customer_Age_bands. invoiceitems_view_by_overall_across_customer_age_band =\ invoiceitems_view.groupby( [], category="customer_Age_band" )

In [ ]:

# Distribution representing the cumulative TotalCost of item, categorized by their respective
# customer_Age_band, over time.
feature_group =\
invoiceitems_view_by_overall_across_customer_age_band.aggregate_over(
    "TotalCost", method=fb.AggFunc.SUM,
    feature_names=[
        "OVERALL_item_TotalCost_across_customer_Age_bands"
        + "_" + w for w in windows
    ],
    windows=windows
)
# Get OVERALL_item_TotalCost_across_customer_Age_bands_28d object from feature group.
overall_item_totalcost_across_customer_age_bands_28d =\
feature_group["OVERALL_item_TotalCost_across_customer_Age_bands_28d"]

# Distribution representing the cumulative TotalCost of item, categorized by their respective # customer_Age_band, over time. feature_group =\ invoiceitems_view_by_overall_across_customer_age_band.aggregate_over( "TotalCost", method=fb.AggFunc.SUM, feature_names=[ "OVERALL_item_TotalCost_across_customer_Age_bands" + "_" + w for w in windows ], windows=windows ) # Get OVERALL_item_TotalCost_across_customer_Age_bands_28d object from feature group. overall_item_totalcost_across_customer_age_bands_28d =\ feature_group["OVERALL_item_TotalCost_across_customer_Age_bands_28d"]

Derive Similarity feature across entities¶

See list of feature transforms

In [ ]:

# Derive Similarity feature from cosine similarity between
# PRODUCT_item_TotalCost_across_customer_Age_bands_28d
# and OVERALL_item_TotalCost_across_customer_Age_bands_28d
product_vs_overall_item_totalcost_across_customer_age_bands_28d =\
product_item_totalcost_across_customer_age_bands_28d.cd.cosine_similarity(
        overall_item_totalcost_across_customer_age_bands_28d
    )
# Give a name to new feature
product_vs_overall_item_totalcost_across_customer_age_bands_28d.name = \
"PRODUCT_vs_OVERALL_item_TotalCost_across_customer_Age_bands_28d"

# Derive Similarity feature from cosine similarity between # PRODUCT_item_TotalCost_across_customer_Age_bands_28d # and OVERALL_item_TotalCost_across_customer_Age_bands_28d product_vs_overall_item_totalcost_across_customer_age_bands_28d =\ product_item_totalcost_across_customer_age_bands_28d.cd.cosine_similarity( overall_item_totalcost_across_customer_age_bands_28d ) # Give a name to new feature product_vs_overall_item_totalcost_across_customer_age_bands_28d.name = \ "PRODUCT_vs_OVERALL_item_TotalCost_across_customer_Age_bands_28d"

Preview feature¶

Read on the feature primary entity concept

Read on the serving entity concept

See SDK reference for observation table

See SDK reference to preview feature

In [ ]:

#Check the primary entity of the feature'
product_vs_overall_item_totalcost_across_customer_age_bands_28d.primary_entity

#Check the primary entity of the feature' product_vs_overall_item_totalcost_across_customer_age_bands_28d.primary_entity

In [ ]:

#Get observation table: 'Preview Table with 10 items'
preview_table = catalog.get_observation_table(
	"Preview Table with 10 items"
)

#Get observation table: 'Preview Table with 10 items' preview_table = catalog.get_observation_table( "Preview Table with 10 items" )

In [ ]:

#Preview PRODUCT_vs_OVERALL_item_TotalCost_across_customer_Age_bands_28d
product_vs_overall_item_totalcost_across_customer_age_bands_28d.preview(
	preview_table
)

#Preview PRODUCT_vs_OVERALL_item_TotalCost_across_customer_Age_bands_28d product_vs_overall_item_totalcost_across_customer_age_bands_28d.preview( preview_table )

Save feature¶

See SDK reference to save feature

In [ ]:

# Save feature
product_vs_overall_item_totalcost_across_customer_age_bands_28d.save()

# Save feature product_vs_overall_item_totalcost_across_customer_age_bands_28d.save()

Add description and see feature definition file¶

See SDK reference for the definition file of a feature

In [ ]:

# Add description
product_vs_overall_item_totalcost_across_customer_age_bands_28d.update_description(
	"Similarity between the product and all products measured by the Cosine"
	" Similarity between the Distribution representing the cumulative "
	"TotalCost of item, categorized by their respective customer_Age_band, "
	"over 28d for both entities."
)
# See feature definition file
product_vs_overall_item_totalcost_across_customer_age_bands_28d.definition

# Add description product_vs_overall_item_totalcost_across_customer_age_bands_28d.update_description( "Similarity between the product and all products measured by the Cosine" " Similarity between the Distribution representing the cumulative " "TotalCost of item, categorized by their respective customer_Age_band, " "over 28d for both entities." ) # See feature definition file product_vs_overall_item_totalcost_across_customer_age_bands_28d.definition