9. Create Window Aggregates from Event Table

Creating Features from the Event Table¶

In this step, we define 21 features by aggregating events from the PRIOR_APPLICATIONS table within a 104-week window, after applying filters on the CONTRACT_STATUS and CONTRACT_TYPE columns.

1. Statistical Aggregations (Avg, Max, Min, Pct)¶

These features summarize prior applications using basic statistical functions:

Average (Avg)¶

CLIENT_Avg_of_Prior_Applications_CNT_PAYMENTs_104w
CLIENT_Avg_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w
CLIENT_Avg_of_Approved_Status_Prior_Applications_CNT_PAYMENTs_104w

Percentage (Pct)¶

CLIENT_Pct_of_Prior_Applications_is_Refused_Status_52w
CLIENT_Pct_of_Prior_Applications_is_Refused_Status_Cash_loans_Contract_type_104w

Maximum (Max)¶

CLIENT_Max_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w
CLIENT_Max_of_Prior_Applications_AMT_CREDIT_To_AMT_APPLICATIONs_104w
CLIENT_Max_of_Prior_Applications_Application-Credit_Gaps_104w
CLIENT_Max_of_Prior_Applications_CNT_PAYMENTs_104w

Minimum (Min)¶

CLIENT_Min_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_AMT_ANNUITY_To_AMT_CREDITs_104w
CLIENT_Min_of_Cash_loans_Contract_type_Prior_Applications_CNT_PAYMENTs_104w

2. Features Based on Latest Events and Time Since Last Event¶

These features track the most recent events and the time elapsed since the last recorded event:

Latest Event Values¶

CLIENT_Latest_Refused_Status_Cash_loans_Contract_type_Prior_Application_YIELD_GROUP_104w
CLIENT_Latest_Revolving_loans_Contract_type_Prior_Application_AMT_APPLICATION_104w

Time Since Latest Event¶

CLIENT_Time_Since_Latest_Prior_Application_DECISION_DATE_104w

3. Features Based on Inter-Event Time¶

These features measure the time intervals between successive prior applications:

CLIENT_Max_of_Time_between_2_Approved_Status_Prior_Applications_for_the_Client_104w
CLIENT_Max_of_Time_between_2_Prior_Applications_for_the_Client_104w

4. Aggregations Across `YIELD_GROUP`¶

These features aggregate prior applications based on the YIELD_GROUP category:

CLIENT_Approved_Status_Prior_Applications_AMT_CREDITs_by_Prior_Application_YIELD_GROUP_104w
CLIENT_Count_of_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w
CLIENT_Count_of_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w

5. Features Derived from Two Other Features¶

These features are computed as ratios or comparisons between two aggregated features:

CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_To_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w
CLIENT_vs_OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w

Activate catalog¶

See SDK reference for catalogs

In [1]:

            
                Copied!
                
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)

16:42:02 | WARNING  | Service endpoint is inaccessible: http://featurebyte-server:8088/
16:42:02 | INFO     | Using profile: tutorial
16:42:02 | INFO     | Using configuration file at: /Users/gxav/.featurebyte/config.yaml
16:42:02 | INFO     | Active profile: tutorial (https://tutorials.featurebyte.com/api/v1)
16:42:02 | INFO     | SDK version: 2.1.0.dev113
16:42:02 | INFO     | No catalog activated.
16:42:02 | INFO     | Catalog activated: Credit Default Dataset SDK Tutorial

Get view from table¶

See SDK reference for views

In [2]:

            
                Copied!
                
# Get view from PRIOR_APPLICATIONS event table.
prior_applications_view = catalog.get_view("PRIOR_APPLICATIONS")
# Get view from PRIOR_APPLICATIONS event table.
prior_applications_view = catalog.get_view("PRIOR_APPLICATIONS")

Create ratio and difference columns¶

See SDK reference for view columns

In [3]:

            
                Copied!
                
prior_applications_view["AMT_ANNUITY To AMT_CREDIT"] = (
    prior_applications_view["AMT_ANNUITY"] / prior_applications_view["AMT_CREDIT"]
)
prior_applications_view["AMT_ANNUITY To AMT_CREDIT"] = (
    prior_applications_view["AMT_ANNUITY"] / prior_applications_view["AMT_CREDIT"]
)

In [4]:

            
                Copied!
                
prior_applications_view["AMT_CREDIT To AMT_APPLICATION"] = (
    prior_applications_view["AMT_CREDIT"] / prior_applications_view["AMT_APPLICATION"]
)
prior_applications_view["AMT_CREDIT To AMT_APPLICATION"] = (
    prior_applications_view["AMT_CREDIT"] / prior_applications_view["AMT_APPLICATION"]
)

In [5]:

            
                Copied!
                
prior_applications_view["Application-Credit Gap"] = (
    prior_applications_view["AMT_APPLICATION"] - prior_applications_view["AMT_CREDIT"]
)
prior_applications_view["Application-Credit Gap"] = (
    prior_applications_view["AMT_APPLICATION"] - prior_applications_view["AMT_CREDIT"]
)

Filter View¶

Read on view subsetting

In [6]:

            
                Copied!
                
# Filter prior_applications_view with CONTRACT_STATUS equal to Approved.
cond = prior_applications_view["CONTRACT_STATUS"] == "Approved"
prior_applications_approved_status_view = prior_applications_view[cond]
# Filter prior_applications_view with CONTRACT_STATUS equal to Approved.
cond = prior_applications_view["CONTRACT_STATUS"] == "Approved"
prior_applications_approved_status_view = prior_applications_view[cond]

In [7]:

            
                Copied!
                
# Filter prior_applications_view with CONTRACT_TYPE equal to Revolving loans.
cond = prior_applications_view["CONTRACT_TYPE"] == "Revolving loans"
prior_applications_revolving_loans_contract_type_view = prior_applications_view[cond]
# Filter prior_applications_view with CONTRACT_TYPE equal to Revolving loans.
cond = prior_applications_view["CONTRACT_TYPE"] == "Revolving loans"
prior_applications_revolving_loans_contract_type_view = prior_applications_view[cond]

In [8]:

            
                Copied!
                
# Filter prior_applications_view with CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_cash_loans_contract_type_view = prior_applications_view[cond]
# Filter prior_applications_view with CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_cash_loans_contract_type_view = prior_applications_view[cond]

In [9]:

            
                Copied!
                
# Filter prior_applications_view with CONTRACT_STATUS equal to Refused.
cond = prior_applications_view["CONTRACT_STATUS"] == "Refused"
prior_applications_view["is Refused Status"] = cond.astype(int)
# Filter prior_applications_view with CONTRACT_STATUS equal to Refused.
cond = prior_applications_view["CONTRACT_STATUS"] == "Refused"
prior_applications_view["is Refused Status"] = cond.astype(int)

In [10]:

            
                Copied!
                
# Filter prior_applications_view with CONTRACT_STATUS equal to Approved and CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_STATUS"] == "Approved"
cond &= prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_approved_status_cash_loans_contract_type_view = prior_applications_view[cond]
# Filter prior_applications_view with CONTRACT_STATUS equal to Approved and CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_STATUS"] == "Approved"
cond &= prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_approved_status_cash_loans_contract_type_view = prior_applications_view[cond]

In [11]:

            
                Copied!
                
# Filter prior_applications_view with CONTRACT_STATUS equal to Refused and CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_STATUS"] == "Refused"
cond &= prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_refused_status_cash_loans_contract_type_view = prior_applications_view[cond]
# Filter prior_applications_view with CONTRACT_STATUS equal to Refused and CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_STATUS"] == "Refused"
cond &= prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_refused_status_cash_loans_contract_type_view = prior_applications_view[cond]

Create Flag column¶

In [12]:

            
                Copied!
                
# Flag CONTRACT_STATUS equal to Refused and CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_STATUS"] == "Refused"
cond &= prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_view["is Refused Status Cash loans Contract_type"] = cond.astype(int)
# Flag CONTRACT_STATUS equal to Refused and CONTRACT_TYPE equal to Cash loans.
cond = prior_applications_view["CONTRACT_STATUS"] == "Refused"
cond &= prior_applications_view["CONTRACT_TYPE"] == "Cash loans"
prior_applications_view["is Refused Status Cash loans Contract_type"] = cond.astype(int)

Derive Inter-Event Time columns¶

See SDK reference to extract lags

In [13]:

            
                Copied!
                
# Extract InterEventTime by Client
prior_applications_view["Time between 2 Prior Applications for the Client"] = (
    prior_applications_view["DECISION_DATE"]
    - prior_applications_view["DECISION_DATE"].lag("CLIENT_ID")
).dt.day
# Extract InterEventTime by Client
prior_applications_view["Time between 2 Prior Applications for the Client"] = (
    prior_applications_view["DECISION_DATE"]
    - prior_applications_view["DECISION_DATE"].lag("CLIENT_ID")
).dt.day

In [14]:

            
                Copied!
                
                    
                    
                
                

        
# Extract InterEventTime by Client
prior_applications_approved_status_view[
    "Time between 2 Approved Status Prior Applications for the Client"
] = (
    prior_applications_approved_status_view["DECISION_DATE"]
    - prior_applications_approved_status_view["DECISION_DATE"].lag("CLIENT_ID")
).dt.day
# Extract InterEventTime by Client
prior_applications_approved_status_view[
    "Time between 2 Approved Status Prior Applications for the Client"
] = (
    prior_applications_approved_status_view["DECISION_DATE"]
    - prior_applications_approved_status_view["DECISION_DATE"].lag("CLIENT_ID")
).dt.day

Do window aggregation from PRIOR_APPLICATIONS¶

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 [15]:

            
                Copied!
                
# Group PRIOR_APPLICATIONS view by Client entity (CLIENT_ID).
prior_applications_view_by_client = prior_applications_view.groupby(["CLIENT_ID"])
# Group PRIOR_APPLICATIONS view by Client entity (CLIENT_ID).
prior_applications_view_by_client = prior_applications_view.groupby(["CLIENT_ID"])

In [16]:

            
                Copied!
                
                    
                    
                
                

        
# Get Latest Prior Application's DECISION_DATE for the Client over a 104w period.
client_latest_prior_application_decision_date_104w = (
    prior_applications_view_by_client.aggregate_over(
        "DECISION_DATE",
        method="latest",
        feature_names=["CLIENT_Latest_Prior_Application_DECISION_DATE_104w"],
        windows=["104w"],
    )["CLIENT_Latest_Prior_Application_DECISION_DATE_104w"]
)
# Get Latest Prior Application's DECISION_DATE for the Client over a 104w period.
client_latest_prior_application_decision_date_104w = (
    prior_applications_view_by_client.aggregate_over(
        "DECISION_DATE",
        method="latest",
        feature_names=["CLIENT_Latest_Prior_Application_DECISION_DATE_104w"],
        windows=["104w"],
    )["CLIENT_Latest_Prior_Application_DECISION_DATE_104w"]
)

In [17]:

            
                Copied!
                
                    
                    
                
                

        
# Get Avg of CNT_PAYMENT for the Client over time.
client_avg_of_prior_applications_cnt_payments_104w = (
    prior_applications_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="avg",
        feature_names=["CLIENT_Avg_of_Prior_Applications_CNT_PAYMENTs_104w"],
        windows=["104w"],
    )["CLIENT_Avg_of_Prior_Applications_CNT_PAYMENTs_104w"]
)
# Get Avg of CNT_PAYMENT for the Client over time.
client_avg_of_prior_applications_cnt_payments_104w = (
    prior_applications_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="avg",
        feature_names=["CLIENT_Avg_of_Prior_Applications_CNT_PAYMENTs_104w"],
        windows=["104w"],
    )["CLIENT_Avg_of_Prior_Applications_CNT_PAYMENTs_104w"]
)

In [18]:

            
                Copied!
                
                    
                    
                
                

        
# Get Max of CNT_PAYMENT for the Client over time.
client_max_of_prior_applications_cnt_payments_104w = (
    prior_applications_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="max",
        feature_names=["CLIENT_Max_of_Prior_Applications_CNT_PAYMENTs_104w"],
        windows=["104w"],
    )["CLIENT_Max_of_Prior_Applications_CNT_PAYMENTs_104w"]
)
# Get Max of CNT_PAYMENT for the Client over time.
client_max_of_prior_applications_cnt_payments_104w = (
    prior_applications_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="max",
        feature_names=["CLIENT_Max_of_Prior_Applications_CNT_PAYMENTs_104w"],
        windows=["104w"],
    )["CLIENT_Max_of_Prior_Applications_CNT_PAYMENTs_104w"]
)

In [21]:

            
                Copied!
                
                    
                    
                
                

        
# Get Max of AMT_CREDIT To AMT_APPLICATION for the Client over time.
client_max_of_prior_applications_amt_credit_to_amt_applications_104w = (
    prior_applications_view_by_client.aggregate_over(
        "AMT_CREDIT To AMT_APPLICATION",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Prior_Applications_AMT_CREDIT_To_AMT_APPLICATIONs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Prior_Applications_AMT_CREDIT_To_AMT_APPLICATIONs_104w"]
)
# Get Max of AMT_CREDIT To AMT_APPLICATION for the Client over time.
client_max_of_prior_applications_amt_credit_to_amt_applications_104w = (
    prior_applications_view_by_client.aggregate_over(
        "AMT_CREDIT To AMT_APPLICATION",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Prior_Applications_AMT_CREDIT_To_AMT_APPLICATIONs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Prior_Applications_AMT_CREDIT_To_AMT_APPLICATIONs_104w"]
)

In [22]:

            
                Copied!
                
                    
                    
                
                

        
# Get Max of Application-Credit Gap for the Client over time.
client_max_of_prior_applications_application_credit_gaps_104w = (
    prior_applications_view_by_client.aggregate_over(
        "Application-Credit Gap",
        method="max",
        feature_names=["CLIENT_Max_of_Prior_Applications_Application-Credit_Gaps_104w"],
        windows=["104w"],
    )["CLIENT_Max_of_Prior_Applications_Application-Credit_Gaps_104w"]
)
# Get Max of Application-Credit Gap for the Client over time.
client_max_of_prior_applications_application_credit_gaps_104w = (
    prior_applications_view_by_client.aggregate_over(
        "Application-Credit Gap",
        method="max",
        feature_names=["CLIENT_Max_of_Prior_Applications_Application-Credit_Gaps_104w"],
        windows=["104w"],
    )["CLIENT_Max_of_Prior_Applications_Application-Credit_Gaps_104w"]
)

In [23]:

            
                Copied!
                
                    
                    
                
                

        
# Get Pct of is Refused Status for the Client over time.
client_pct_of_prior_applications_is_refused_status_52w = (
    prior_applications_view_by_client.aggregate_over(
        "is Refused Status",
        method="avg",
        feature_names=["CLIENT_Pct_of_Prior_Applications_is_Refused_Status_52w"],
        windows=["52w"],
    )["CLIENT_Pct_of_Prior_Applications_is_Refused_Status_52w"]
)
# Get Pct of is Refused Status for the Client over time.
client_pct_of_prior_applications_is_refused_status_52w = (
    prior_applications_view_by_client.aggregate_over(
        "is Refused Status",
        method="avg",
        feature_names=["CLIENT_Pct_of_Prior_Applications_is_Refused_Status_52w"],
        windows=["52w"],
    )["CLIENT_Pct_of_Prior_Applications_is_Refused_Status_52w"]
)

In [24]:

            
                Copied!
                
                    
                    
                
                

        
# Get Pct of is Refused Status Cash loans Contract_type for the Client over time.
client_pct_of_prior_applications_is_refused_status_cash_loans_contract_type_104w = (
    prior_applications_view_by_client.aggregate_over(
        "is Refused Status Cash loans Contract_type",
        method="avg",
        feature_names=[
            "CLIENT_Pct_of_Prior_Applications_is_Refused_Status_Cash_loans_Contract_type_104w"
        ],
        windows=["104w"],
    )["CLIENT_Pct_of_Prior_Applications_is_Refused_Status_Cash_loans_Contract_type_104w"]
)
# Get Pct of is Refused Status Cash loans Contract_type for the Client over time.
client_pct_of_prior_applications_is_refused_status_cash_loans_contract_type_104w = (
    prior_applications_view_by_client.aggregate_over(
        "is Refused Status Cash loans Contract_type",
        method="avg",
        feature_names=[
            "CLIENT_Pct_of_Prior_Applications_is_Refused_Status_Cash_loans_Contract_type_104w"
        ],
        windows=["104w"],
    )["CLIENT_Pct_of_Prior_Applications_is_Refused_Status_Cash_loans_Contract_type_104w"]
)

In [ ]:

            
                Copied!
                
# Group PRIOR_APPLICATIONS view by Client entity (CLIENT_ID) across different YIELD_GROUPs.
prior_applications_view_by_client_across_yield_groups = prior_applications_view.groupby(
    ["CLIENT_ID"], category="YIELD_GROUP"
)
# Group PRIOR_APPLICATIONS view by Client entity (CLIENT_ID) across different YIELD_GROUPs.
prior_applications_view_by_client_across_yield_groups = prior_applications_view.groupby(
    ["CLIENT_ID"], category="YIELD_GROUP"
)

In [ ]:

            
                Copied!
                
                    
                    
                
                

        
# Count Prior Applications across different YIELD_GROUPs for the Client over time.
client_count_of_prior_applications_by_prior_application_yield_group_104w = (
    prior_applications_view_by_client_across_yield_groups.aggregate_over(
        None,
        method="count",
        feature_names=[
            "CLIENT_Count_of_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Count_of_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"]
)
# Count Prior Applications across different YIELD_GROUPs for the Client over time.
client_count_of_prior_applications_by_prior_application_yield_group_104w = (
    prior_applications_view_by_client_across_yield_groups.aggregate_over(
        None,
        method="count",
        feature_names=[
            "CLIENT_Count_of_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Count_of_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"]
)

Do window aggregation from prior_applications_approved_status_cash_loans_contract_type_view¶

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 [25]:

            
                Copied!
                
                    
                    
                
                

        
# Group prior_applications_approved_status_cash_loans_contract_type_view view by Client entity (CLIENT_ID).
prior_applications_approved_status_cash_loans_contract_type_view_by_client = (
    prior_applications_approved_status_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"]
    )
)
# Group prior_applications_approved_status_cash_loans_contract_type_view view by Client entity (CLIENT_ID).
prior_applications_approved_status_cash_loans_contract_type_view_by_client = (
    prior_applications_approved_status_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"]
    )
)

In [ ]:

            
                Copied!
                
                    
                    
                
                

        
# Get Min of Approved Status Cash loans Contract_type AMT_ANNUITY To AMT_CREDIT for the Client over
# time.
client_min_of_approved_status_cash_loans_contract_type_prior_applications_amt_annuity_to_amt_credits_104w = (
    prior_applications_approved_status_cash_loans_contract_type_view_by_client.aggregate_over(
        "AMT_ANNUITY To AMT_CREDIT",
        method="min",
        feature_names=[
            "CLIENT_Min_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_AMT_ANNUITY_To_AMT_CREDITs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Min_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_AMT_ANNUITY_To_AMT_CREDITs_104w"]
)
# Get Min of Approved Status Cash loans Contract_type AMT_ANNUITY To AMT_CREDIT for the Client over
# time.
client_min_of_approved_status_cash_loans_contract_type_prior_applications_amt_annuity_to_amt_credits_104w = (
    prior_applications_approved_status_cash_loans_contract_type_view_by_client.aggregate_over(
        "AMT_ANNUITY To AMT_CREDIT",
        method="min",
        feature_names=[
            "CLIENT_Min_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_AMT_ANNUITY_To_AMT_CREDITs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Min_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_AMT_ANNUITY_To_AMT_CREDITs_104w"]
)

In [26]:

            
                Copied!
                
                    
                    
                
                

        
# Group prior_applications_approved_status_cash_loans_contract_type_view view by Client entity
# (CLIENT_ID) across different PRODUCT_COMBINATIONs.
prior_applications_approved_status_cash_loans_contract_type_view_by_client_across_product_combinations = (
    prior_applications_approved_status_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"], category="PRODUCT_COMBINATION"
    )
)
# Group prior_applications_approved_status_cash_loans_contract_type_view view by Client entity
# (CLIENT_ID) across different PRODUCT_COMBINATIONs.
prior_applications_approved_status_cash_loans_contract_type_view_by_client_across_product_combinations = (
    prior_applications_approved_status_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"], category="PRODUCT_COMBINATION"
    )
)

In [27]:

            
                Copied!
                
                    
                    
                
                

        
# Count Approved Status Cash loans Contract_type Prior Applications across different
# PRODUCT_COMBINATIONs for the Client over time.
client_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w = (
    prior_applications_approved_status_cash_loans_contract_type_view_by_client_across_product_combinations.aggregate_over(
        None,
        method="count",
        feature_names=[
            "CLIENT_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"
        ],
        windows=["104w"],
    )["CLIENT_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"]
)
# Count Approved Status Cash loans Contract_type Prior Applications across different
# PRODUCT_COMBINATIONs for the Client over time.
client_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w = (
    prior_applications_approved_status_cash_loans_contract_type_view_by_client_across_product_combinations.aggregate_over(
        None,
        method="count",
        feature_names=[
            "CLIENT_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"
        ],
        windows=["104w"],
    )["CLIENT_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"]
)

In [29]:

            
                Copied!
                
                    
                    
                
                

        
# Group prior_applications_approved_status_cash_loans_contract_type_view view across different
# PRODUCT_COMBINATIONs.
prior_applications_approved_status_cash_loans_contract_type_view_by_overall_across_product_combinations = (
    prior_applications_approved_status_cash_loans_contract_type_view.groupby(
        [], category="PRODUCT_COMBINATION"
    )
)
# Group prior_applications_approved_status_cash_loans_contract_type_view view across different
# PRODUCT_COMBINATIONs.
prior_applications_approved_status_cash_loans_contract_type_view_by_overall_across_product_combinations = (
    prior_applications_approved_status_cash_loans_contract_type_view.groupby(
        [], category="PRODUCT_COMBINATION"
    )
)

In [30]:

            
                Copied!
                
                    
                    
                
                

        
# Count Approved Status Cash loans Contract_type Prior Applications across different
# PRODUCT_COMBINATIONs over time.
overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w = (
    prior_applications_approved_status_cash_loans_contract_type_view_by_overall_across_product_combinations.aggregate_over(
        None,
        method="count",
        feature_names=[
            "OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"
        ],
        windows=["104w"],
    )["OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"]
)
# Count Approved Status Cash loans Contract_type Prior Applications across different
# PRODUCT_COMBINATIONs over time.
overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w = (
    prior_applications_approved_status_cash_loans_contract_type_view_by_overall_across_product_combinations.aggregate_over(
        None,
        method="count",
        feature_names=[
            "OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"
        ],
        windows=["104w"],
    )["OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"]
)

Do window aggregation from prior_applications_approved_status_view¶

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 [31]:

            
                Copied!
                
# Group prior_applications_approved_status_view view by Client entity (CLIENT_ID).
prior_applications_approved_status_view_by_client = (
    prior_applications_approved_status_view.groupby(["CLIENT_ID"])
)
# Group prior_applications_approved_status_view view by Client entity (CLIENT_ID).
prior_applications_approved_status_view_by_client = (
    prior_applications_approved_status_view.groupby(["CLIENT_ID"])
)

In [32]:

            
                Copied!
                
                    
                    
                
                

        
# Get Avg of Approved Status AMT_ANNUITY for the Client over time.
client_avg_of_approved_status_prior_applications_amt_annuitys_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_ANNUITY",
        method="avg",
        feature_names=[
            "CLIENT_Avg_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Avg_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"]
)
# Get Avg of Approved Status AMT_ANNUITY for the Client over time.
client_avg_of_approved_status_prior_applications_amt_annuitys_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_ANNUITY",
        method="avg",
        feature_names=[
            "CLIENT_Avg_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Avg_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"]
)

In [33]:

            
                Copied!
                
                    
                    
                
                

        
# Get Max of Approved Status AMT_ANNUITY for the Client over time.
client_max_of_approved_status_prior_applications_amt_annuitys_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_ANNUITY",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"]
)
# Get Max of Approved Status AMT_ANNUITY for the Client over time.
client_max_of_approved_status_prior_applications_amt_annuitys_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_ANNUITY",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"]
)

In [34]:

            
                Copied!
                
                    
                    
                
                

        
# Get Sum of Approved Status AMT_ANNUITY for the Client over time.
client_sum_of_approved_status_prior_applications_amt_annuitys_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_ANNUITY",
        method="sum",
        feature_names=[
            "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"]
)
# Get Sum of Approved Status AMT_ANNUITY for the Client over time.
client_sum_of_approved_status_prior_applications_amt_annuitys_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_ANNUITY",
        method="sum",
        feature_names=[
            "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w"]
)

In [35]:

            
                Copied!
                
                    
                    
                
                

        
# Get Sum of Approved Status AMT_CREDIT for the Client over time.
client_sum_of_approved_status_prior_applications_amt_credits_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_CREDIT",
        method="sum",
        feature_names=[
            "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"]
)
# Get Sum of Approved Status AMT_CREDIT for the Client over time.
client_sum_of_approved_status_prior_applications_amt_credits_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "AMT_CREDIT",
        method="sum",
        feature_names=[
            "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"]
)

In [36]:

            
                Copied!
                
                    
                    
                
                

        
# Get Avg of Approved Status CNT_PAYMENT for the Client over time.
client_avg_of_approved_status_prior_applications_cnt_payments_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="avg",
        feature_names=[
            "CLIENT_Avg_of_Approved_Status_Prior_Applications_CNT_PAYMENTs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Avg_of_Approved_Status_Prior_Applications_CNT_PAYMENTs_104w"]
)
# Get Avg of Approved Status CNT_PAYMENT for the Client over time.
client_avg_of_approved_status_prior_applications_cnt_payments_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="avg",
        feature_names=[
            "CLIENT_Avg_of_Approved_Status_Prior_Applications_CNT_PAYMENTs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Avg_of_Approved_Status_Prior_Applications_CNT_PAYMENTs_104w"]
)

In [37]:

            
                Copied!
                
                    
                    
                
                

        
# Group prior_applications_approved_status_view view by Client entity (CLIENT_ID) across different
# YIELD_GROUPs.
prior_applications_approved_status_view_by_client_across_yield_groups = (
    prior_applications_approved_status_view.groupby(
        ["CLIENT_ID"], category="YIELD_GROUP"
    )
)
# Group prior_applications_approved_status_view view by Client entity (CLIENT_ID) across different
# YIELD_GROUPs.
prior_applications_approved_status_view_by_client_across_yield_groups = (
    prior_applications_approved_status_view.groupby(
        ["CLIENT_ID"], category="YIELD_GROUP"
    )
)

In [38]:

            
                Copied!
                
                    
                    
                
                

        
# Distribution of the total AMT_CREDITs of Approved Status Prior Applications, segmented by Prior
# Application's YIELD_GROUP for the Client over time.
client_approved_status_prior_applications_amt_credits_by_prior_application_yield_group_104w = (
    prior_applications_approved_status_view_by_client_across_yield_groups.aggregate_over(
        "AMT_CREDIT",
        method="sum",
        feature_names=[
            "CLIENT_Approved_Status_Prior_Applications_AMT_CREDITs_by_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Approved_Status_Prior_Applications_AMT_CREDITs_by_Prior_Application_YIELD_GROUP_104w"]
)
# Distribution of the total AMT_CREDITs of Approved Status Prior Applications, segmented by Prior
# Application's YIELD_GROUP for the Client over time.
client_approved_status_prior_applications_amt_credits_by_prior_application_yield_group_104w = (
    prior_applications_approved_status_view_by_client_across_yield_groups.aggregate_over(
        "AMT_CREDIT",
        method="sum",
        feature_names=[
            "CLIENT_Approved_Status_Prior_Applications_AMT_CREDITs_by_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Approved_Status_Prior_Applications_AMT_CREDITs_by_Prior_Application_YIELD_GROUP_104w"]
)

Do window aggregation from prior_applications_refused_status_cash_loans_contract_type_view¶

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 [39]:

            
                Copied!
                
                    
                    
                
                

        
# Group prior_applications_refused_status_cash_loans_contract_type_view view by Client entity
# (CLIENT_ID).
prior_applications_refused_status_cash_loans_contract_type_view_by_client = (
    prior_applications_refused_status_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"]
    )
)
# Group prior_applications_refused_status_cash_loans_contract_type_view view by Client entity
# (CLIENT_ID).
prior_applications_refused_status_cash_loans_contract_type_view_by_client = (
    prior_applications_refused_status_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"]
    )
)

In [40]:

            
                Copied!
                
                    
                    
                
                

        
# Get Latest Refused Status Cash loans Contract_type Prior Application's YIELD_GROUP for the Client
# over a 104w period.
client_latest_refused_status_cash_loans_contract_type_prior_application_yield_group_104w = (
    prior_applications_refused_status_cash_loans_contract_type_view_by_client.aggregate_over(
        "YIELD_GROUP",
        method="latest",
        feature_names=[
            "CLIENT_Latest_Refused_Status_Cash_loans_Contract_type_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Latest_Refused_Status_Cash_loans_Contract_type_Prior_Application_YIELD_GROUP_104w"]
)
# Get Latest Refused Status Cash loans Contract_type Prior Application's YIELD_GROUP for the Client
# over a 104w period.
client_latest_refused_status_cash_loans_contract_type_prior_application_yield_group_104w = (
    prior_applications_refused_status_cash_loans_contract_type_view_by_client.aggregate_over(
        "YIELD_GROUP",
        method="latest",
        feature_names=[
            "CLIENT_Latest_Refused_Status_Cash_loans_Contract_type_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Latest_Refused_Status_Cash_loans_Contract_type_Prior_Application_YIELD_GROUP_104w"]
)

Do window aggregation from prior_applications_revolving_loans_contract_type_view¶

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 [41]:

            
                Copied!
                
# Group prior_applications_revolving_loans_contract_type_view view by Client entity (CLIENT_ID).
prior_applications_revolving_loans_contract_type_view_by_client = (
    prior_applications_revolving_loans_contract_type_view.groupby(["CLIENT_ID"])
)
# Group prior_applications_revolving_loans_contract_type_view view by Client entity (CLIENT_ID).
prior_applications_revolving_loans_contract_type_view_by_client = (
    prior_applications_revolving_loans_contract_type_view.groupby(["CLIENT_ID"])
)

In [42]:

            
                Copied!
                
                    
                    
                
                

        
# Get Latest Revolving loans Contract_type Prior Application's AMT_APPLICATION for the Client over
# a 104w period.
client_latest_revolving_loans_contract_type_prior_application_amt_application_104w = (
    prior_applications_revolving_loans_contract_type_view_by_client.aggregate_over(
        "AMT_APPLICATION",
        method="latest",
        feature_names=[
            "CLIENT_Latest_Revolving_loans_Contract_type_Prior_Application_AMT_APPLICATION_104w"
        ],
        windows=["104w"],
    )["CLIENT_Latest_Revolving_loans_Contract_type_Prior_Application_AMT_APPLICATION_104w"]
)
# Get Latest Revolving loans Contract_type Prior Application's AMT_APPLICATION for the Client over
# a 104w period.
client_latest_revolving_loans_contract_type_prior_application_amt_application_104w = (
    prior_applications_revolving_loans_contract_type_view_by_client.aggregate_over(
        "AMT_APPLICATION",
        method="latest",
        feature_names=[
            "CLIENT_Latest_Revolving_loans_Contract_type_Prior_Application_AMT_APPLICATION_104w"
        ],
        windows=["104w"],
    )["CLIENT_Latest_Revolving_loans_Contract_type_Prior_Application_AMT_APPLICATION_104w"]
)

Do window aggregation from prior_applications_cash_loans_contract_type_view¶

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 [43]:

            
                Copied!
                
# Group prior_applications_cash_loans_contract_type_view view by Client entity (CLIENT_ID).
prior_applications_cash_loans_contract_type_view_by_client = (
    prior_applications_cash_loans_contract_type_view.groupby(["CLIENT_ID"])
)
# Group prior_applications_cash_loans_contract_type_view view by Client entity (CLIENT_ID).
prior_applications_cash_loans_contract_type_view_by_client = (
    prior_applications_cash_loans_contract_type_view.groupby(["CLIENT_ID"])
)

In [44]:

            
                Copied!
                
                    
                    
                
                

        
# Get Min of Cash loans Contract_type CNT_PAYMENT for the Client over time.
client_min_of_cash_loans_contract_type_prior_applications_cnt_payments_104w = (
    prior_applications_cash_loans_contract_type_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="min",
        feature_names=[
            "CLIENT_Min_of_Cash_loans_Contract_type_Prior_Applications_CNT_PAYMENTs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Min_of_Cash_loans_Contract_type_Prior_Applications_CNT_PAYMENTs_104w"]
)
# Get Min of Cash loans Contract_type CNT_PAYMENT for the Client over time.
client_min_of_cash_loans_contract_type_prior_applications_cnt_payments_104w = (
    prior_applications_cash_loans_contract_type_view_by_client.aggregate_over(
        "CNT_PAYMENT",
        method="min",
        feature_names=[
            "CLIENT_Min_of_Cash_loans_Contract_type_Prior_Applications_CNT_PAYMENTs_104w"
        ],
        windows=["104w"],
    )["CLIENT_Min_of_Cash_loans_Contract_type_Prior_Applications_CNT_PAYMENTs_104w"]
)

In [45]:

            
                Copied!
                
                    
                    
                
                

        
# Group prior_applications_cash_loans_contract_type_view view by Client entity (CLIENT_ID) across
# different YIELD_GROUPs.
prior_applications_cash_loans_contract_type_view_by_client_across_yield_groups = (
    prior_applications_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"], category="YIELD_GROUP"
    )
)
# Group prior_applications_cash_loans_contract_type_view view by Client entity (CLIENT_ID) across
# different YIELD_GROUPs.
prior_applications_cash_loans_contract_type_view_by_client_across_yield_groups = (
    prior_applications_cash_loans_contract_type_view.groupby(
        ["CLIENT_ID"], category="YIELD_GROUP"
    )
)

In [46]:

            
                Copied!
                
                    
                    
                
                

        
# Count Cash loans Contract_type Prior Applications across different YIELD_GROUPs for the Client
# over time.
client_count_of_cash_loans_contract_type_prior_applications_by_prior_application_yield_group_104w = (
    prior_applications_cash_loans_contract_type_view_by_client_across_yield_groups.aggregate_over(
        None,
        method="count",
        feature_names=[
            "CLIENT_Count_of_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Count_of_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"]
)
# Count Cash loans Contract_type Prior Applications across different YIELD_GROUPs for the Client
# over time.
client_count_of_cash_loans_contract_type_prior_applications_by_prior_application_yield_group_104w = (
    prior_applications_cash_loans_contract_type_view_by_client_across_yield_groups.aggregate_over(
        None,
        method="count",
        feature_names=[
            "CLIENT_Count_of_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"
        ],
        windows=["104w"],
    )["CLIENT_Count_of_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_YIELD_GROUP_104w"]
)

Derive Time Since feature from latest timestamp¶

See SDK reference to derive time since features

In [47]:

            
                Copied!
                
                    
                    
                
                

        
# Compare CLIENT_Latest_Prior_Application_DECISION_DATE_104w with point-in-time in days
client_time_since_latest_prior_application_decision_date_104w = (
    fb.RequestColumn.point_in_time()
    - client_latest_prior_application_decision_date_104w
).dt.day

# Give a name to new feature
client_time_since_latest_prior_application_decision_date_104w.name = (
    "CLIENT_Time_Since_Latest_Prior_Application_DECISION_DATE_104w"
)
# Compare CLIENT_Latest_Prior_Application_DECISION_DATE_104w with point-in-time in days
client_time_since_latest_prior_application_decision_date_104w = (
    fb.RequestColumn.point_in_time()
    - client_latest_prior_application_decision_date_104w
).dt.day

# Give a name to new feature
client_time_since_latest_prior_application_decision_date_104w.name = (
    "CLIENT_Time_Since_Latest_Prior_Application_DECISION_DATE_104w"
)

Derive A Ratio from two features¶

See list of feature transforms

In [48]:

            
                Copied!
                
                    
                    
                
                

        
# Get the Ratio of CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w To
# CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w
client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w = (
    client_sum_of_approved_status_prior_applications_amt_annuitys_104w
    / client_sum_of_approved_status_prior_applications_amt_credits_104w
)

# Give a name to new feature
client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w.name = (
    "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_To_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"
)
# Get the Ratio of CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w To
# CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w
client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w = (
    client_sum_of_approved_status_prior_applications_amt_annuitys_104w
    / client_sum_of_approved_status_prior_applications_amt_credits_104w
)

# Give a name to new feature
client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w.name = (
    "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_To_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"
)

Derive Similarity feature across entities¶

See list of feature transforms

In [49]:

            
                Copied!
                
                    
                    
                
                

        
# Derive Similarity feature from cosine similarity between
# CLIENT_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w
# and OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w
client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w = (
    client_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w.cd.cosine_similarity(
        overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w
    )
)

# Give a name to new feature
client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w.name = (
    "CLIENT_vs_OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"
)
# Derive Similarity feature from cosine similarity between
# CLIENT_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w
# and OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w
client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w = (
    client_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w.cd.cosine_similarity(
        overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w
    )
)

# Give a name to new feature
client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w.name = (
    "CLIENT_vs_OVERALL_Count_of_Approved_Status_Cash_loans_Contract_type_Prior_Applications_by_Prior_Application_PRODUCT_COMBINATION_104w"
)

Create feature from Inter-Event Time¶

In [50]:

            
                Copied!
                
                    
                    
                
                

        
# Get Max of Time between 2 Prior Applications for the Client over time.
client_max_of_time_between_2_prior_applications_for_the_client_104w = (
    prior_applications_view_by_client.aggregate_over(
        "Time between 2 Prior Applications for the Client",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Time_between_2_Prior_Applications_for_the_Client_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Time_between_2_Prior_Applications_for_the_Client_104w"]
)
# Get Max of Time between 2 Prior Applications for the Client over time.
client_max_of_time_between_2_prior_applications_for_the_client_104w = (
    prior_applications_view_by_client.aggregate_over(
        "Time between 2 Prior Applications for the Client",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Time_between_2_Prior_Applications_for_the_Client_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Time_between_2_Prior_Applications_for_the_Client_104w"]
)

In [51]:

            
                Copied!
                
                    
                    
                
                

        
# Get Max of Time between 2 Approved Status Prior Applications for the Client over time.
client_max_of_time_between_2_approved_status_prior_applications_for_the_client_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "Time between 2 Approved Status Prior Applications for the Client",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Time_between_2_Approved_Status_Prior_Applications_for_the_Client_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Time_between_2_Approved_Status_Prior_Applications_for_the_Client_104w"]
)
# Get Max of Time between 2 Approved Status Prior Applications for the Client over time.
client_max_of_time_between_2_approved_status_prior_applications_for_the_client_104w = (
    prior_applications_approved_status_view_by_client.aggregate_over(
        "Time between 2 Approved Status Prior Applications for the Client",
        method="max",
        feature_names=[
            "CLIENT_Max_of_Time_between_2_Approved_Status_Prior_Applications_for_the_Client_104w"
        ],
        windows=["104w"],
    )["CLIENT_Max_of_Time_between_2_Approved_Status_Prior_Applications_for_the_Client_104w"]
)

Save features¶

See SDK reference to save feature

See SDK reference to save feature group

In [52]:

            
                Copied!
                
                    
                    
                
                

        
fb.FeatureGroup(
    [
        client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w,
        client_time_since_latest_prior_application_decision_date_104w,
        client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w,
        client_pct_of_prior_applications_is_refused_status_cash_loans_contract_type_104w,
        client_pct_of_prior_applications_is_refused_status_52w,
        client_max_of_prior_applications_cnt_payments_104w,
        client_max_of_prior_applications_application_credit_gaps_104w,
        client_max_of_prior_applications_amt_credit_to_amt_applications_104w,
        client_min_of_cash_loans_contract_type_prior_applications_cnt_payments_104w,
        client_max_of_approved_status_prior_applications_amt_annuitys_104w,
        client_min_of_approved_status_cash_loans_contract_type_prior_applications_amt_annuity_to_amt_credits_104w,
        client_latest_revolving_loans_contract_type_prior_application_amt_application_104w,
        client_max_of_time_between_2_prior_applications_for_the_client_104w,
        client_max_of_time_between_2_approved_status_prior_applications_for_the_client_104w,
        client_latest_refused_status_cash_loans_contract_type_prior_application_yield_group_104w,
        client_count_of_prior_applications_by_prior_application_yield_group_104w,
        client_count_of_cash_loans_contract_type_prior_applications_by_prior_application_yield_group_104w,
        client_avg_of_prior_applications_cnt_payments_104w,
        client_avg_of_approved_status_prior_applications_cnt_payments_104w,
        client_avg_of_approved_status_prior_applications_amt_annuitys_104w,
        client_approved_status_prior_applications_amt_credits_by_prior_application_yield_group_104w,
    ]
).save()
fb.FeatureGroup(
    [
        client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w,
        client_time_since_latest_prior_application_decision_date_104w,
        client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w,
        client_pct_of_prior_applications_is_refused_status_cash_loans_contract_type_104w,
        client_pct_of_prior_applications_is_refused_status_52w,
        client_max_of_prior_applications_cnt_payments_104w,
        client_max_of_prior_applications_application_credit_gaps_104w,
        client_max_of_prior_applications_amt_credit_to_amt_applications_104w,
        client_min_of_cash_loans_contract_type_prior_applications_cnt_payments_104w,
        client_max_of_approved_status_prior_applications_amt_annuitys_104w,
        client_min_of_approved_status_cash_loans_contract_type_prior_applications_amt_annuity_to_amt_credits_104w,
        client_latest_revolving_loans_contract_type_prior_application_amt_application_104w,
        client_max_of_time_between_2_prior_applications_for_the_client_104w,
        client_max_of_time_between_2_approved_status_prior_applications_for_the_client_104w,
        client_latest_refused_status_cash_loans_contract_type_prior_application_yield_group_104w,
        client_count_of_prior_applications_by_prior_application_yield_group_104w,
        client_count_of_cash_loans_contract_type_prior_applications_by_prior_application_yield_group_104w,
        client_avg_of_prior_applications_cnt_payments_104w,
        client_avg_of_approved_status_prior_applications_cnt_payments_104w,
        client_avg_of_approved_status_prior_applications_amt_annuitys_104w,
        client_approved_status_prior_applications_amt_credits_by_prior_application_yield_group_104w,
    ]
).save()

Done! |████████████████████████████████████████| 100% in 12.2s (0.08%/s)        
Done! |████████████████████████████████████████| 100% in 6.1s (0.17%/s)         
Loading Feature(s) |████████████████████████████████████████| 21/21 [100%] in 0.

Add description¶

In [53]:

            
                Copied!
                
                    
                    
                
                

        
client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w.update_description(
    "Similarity between an individual Client and other Clients by "
    "calculating the Cosine Similarity between the Count of Approved Status"
    " Cash loans Contract_type Prior Applications across different Prior "
    "Application PRODUCT_COMBINATION categories for both the Client and all"
    " Clients over a 104w period."
)
client_time_since_latest_prior_application_decision_date_104w.update_description(
    "Time (in days) Since Latest Prior Application's DECISION_DATE for the "
    "Client over a 104w period."
)
client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w.update_description(
    "Ratio of "
    "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w To "
    "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"
)
client_pct_of_prior_applications_is_refused_status_cash_loans_contract_type_104w.update_description(
    "Pct of Prior Applications is Refused Status Cash loans Contract_type "
    "for the Client over a 104w period."
)
client_pct_of_prior_applications_is_refused_status_52w.update_description(
    "Pct of Prior Applications is Refused Status for the Client over a 52w " "period."
)
client_max_of_prior_applications_cnt_payments_104w.update_description(
    "Max of Prior Applications CNT_PAYMENTs for the Client over a 104w " "period."
)
client_max_of_prior_applications_application_credit_gaps_104w.update_description(
    "Max of Prior Applications Application-Credit Gaps for the Client over "
    "a 104w period."
)
client_max_of_prior_applications_amt_credit_to_amt_applications_104w.update_description(
    "Max of Prior Applications AMT_CREDIT To AMT_APPLICATIONs for the "
    "Client over a 104w period."
)
client_min_of_cash_loans_contract_type_prior_applications_cnt_payments_104w.update_description(
    "Min of Cash loans Contract_type Prior Applications CNT_PAYMENTs for "
    "the Client over a 104w period."
)
client_max_of_approved_status_prior_applications_amt_annuitys_104w.update_description(
    "Max of Approved Status Prior Applications AMT_ANNUITYs for the Client "
    "over a 104w period."
)
client_min_of_approved_status_cash_loans_contract_type_prior_applications_amt_annuity_to_amt_credits_104w.update_description(
    "Min of Approved Status Cash loans Contract_type Prior Applications "
    "AMT_ANNUITY To AMT_CREDITs for the Client over a 104w period."
)
client_latest_revolving_loans_contract_type_prior_application_amt_application_104w.update_description(
    "Latest Revolving loans Contract_type Prior Application's "
    "AMT_APPLICATION for the Client over a 104w period."
)
client_max_of_time_between_2_prior_applications_for_the_client_104w.update_description(
    "Max of Time between 2 Prior Applications for the Client over a 104w " "period."
)
client_max_of_time_between_2_approved_status_prior_applications_for_the_client_104w.update_description(
    "Max of Time between 2 Approved Status Prior Applications for the "
    "Client over a 104w period."
)
client_latest_refused_status_cash_loans_contract_type_prior_application_yield_group_104w.update_description(
    "Latest Refused Status Cash loans Contract_type Prior Application's "
    "YIELD_GROUP for the Client over a 104w period."
)
client_count_of_prior_applications_by_prior_application_yield_group_104w.update_description(
    "Count of Prior Applications across different Prior Application "
    "YIELD_GROUP categories for the Client over a 104w period. It provides "
    "a detailed breakdown in the form of a dictionary, where each key "
    "represents a unique YIELD_GROUP. The corresponding value for each key "
    "is the count within that YIELD_GROUP. This distribution offers "
    "insights into the spread across different YIELD_GROUPs for the Client."
)
client_count_of_cash_loans_contract_type_prior_applications_by_prior_application_yield_group_104w.update_description(
    "Count of Cash loans Contract_type Prior Applications across different "
    "Prior Application YIELD_GROUP categories for the Client over a 104w "
    "period. It provides a detailed breakdown in the form of a dictionary, "
    "where each key represents a unique YIELD_GROUP. The corresponding "
    "value for each key is the count within that YIELD_GROUP. This "
    "distribution offers insights into the spread across different "
    "YIELD_GROUPs for the Client."
)
client_avg_of_prior_applications_cnt_payments_104w.update_description(
    "Avg of Prior Applications CNT_PAYMENTs for the Client over a 104w " "period."
)
client_avg_of_approved_status_prior_applications_cnt_payments_104w.update_description(
    "Avg of Approved Status Prior Applications CNT_PAYMENTs for the Client "
    "over a 104w period."
)
client_avg_of_approved_status_prior_applications_amt_annuitys_104w.update_description(
    "Avg of Approved Status Prior Applications AMT_ANNUITYs for the Client "
    "over a 104w period."
)
client_approved_status_prior_applications_amt_credits_by_prior_application_yield_group_104w.update_description(
    "Distribution of the total AMT_CREDITs of Approved Status Prior "
    "Applications, segmented by Prior Application's YIELD_GROUP for the "
    "Client over a 104w period. It provides a detailed breakdown in the "
    "form of a dictionary, where each key represents a unique YIELD_GROUP. "
    "The corresponding value for each key is the cumulative sum of "
    "AMT_CREDITs within that YIELD_GROUP. This distribution offers insights"
    " into the AMT_CREDIT spread across different YIELD_GROUPs for the "
    "Client."
)
client_vs_overall_count_of_approved_status_cash_loans_contract_type_prior_applications_by_prior_application_product_combination_104w.update_description(
    "Similarity between an individual Client and other Clients by "
    "calculating the Cosine Similarity between the Count of Approved Status"
    " Cash loans Contract_type Prior Applications across different Prior "
    "Application PRODUCT_COMBINATION categories for both the Client and all"
    " Clients over a 104w period."
)
client_time_since_latest_prior_application_decision_date_104w.update_description(
    "Time (in days) Since Latest Prior Application's DECISION_DATE for the "
    "Client over a 104w period."
)
client_sum_of_approved_status_prior_applications_amt_annuitys_to_sum_of_approved_status_prior_applications_amt_credits_104w.update_description(
    "Ratio of "
    "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_ANNUITYs_104w To "
    "CLIENT_Sum_of_Approved_Status_Prior_Applications_AMT_CREDITs_104w"
)
client_pct_of_prior_applications_is_refused_status_cash_loans_contract_type_104w.update_description(
    "Pct of Prior Applications is Refused Status Cash loans Contract_type "
    "for the Client over a 104w period."
)
client_pct_of_prior_applications_is_refused_status_52w.update_description(
    "Pct of Prior Applications is Refused Status for the Client over a 52w " "period."
)
client_max_of_prior_applications_cnt_payments_104w.update_description(
    "Max of Prior Applications CNT_PAYMENTs for the Client over a 104w " "period."
)
client_max_of_prior_applications_application_credit_gaps_104w.update_description(
    "Max of Prior Applications Application-Credit Gaps for the Client over "
    "a 104w period."
)
client_max_of_prior_applications_amt_credit_to_amt_applications_104w.update_description(
    "Max of Prior Applications AMT_CREDIT To AMT_APPLICATIONs for the "
    "Client over a 104w period."
)
client_min_of_cash_loans_contract_type_prior_applications_cnt_payments_104w.update_description(
    "Min of Cash loans Contract_type Prior Applications CNT_PAYMENTs for "
    "the Client over a 104w period."
)
client_max_of_approved_status_prior_applications_amt_annuitys_104w.update_description(
    "Max of Approved Status Prior Applications AMT_ANNUITYs for the Client "
    "over a 104w period."
)
client_min_of_approved_status_cash_loans_contract_type_prior_applications_amt_annuity_to_amt_credits_104w.update_description(
    "Min of Approved Status Cash loans Contract_type Prior Applications "
    "AMT_ANNUITY To AMT_CREDITs for the Client over a 104w period."
)
client_latest_revolving_loans_contract_type_prior_application_amt_application_104w.update_description(
    "Latest Revolving loans Contract_type Prior Application's "
    "AMT_APPLICATION for the Client over a 104w period."
)
client_max_of_time_between_2_prior_applications_for_the_client_104w.update_description(
    "Max of Time between 2 Prior Applications for the Client over a 104w " "period."
)
client_max_of_time_between_2_approved_status_prior_applications_for_the_client_104w.update_description(
    "Max of Time between 2 Approved Status Prior Applications for the "
    "Client over a 104w period."
)
client_latest_refused_status_cash_loans_contract_type_prior_application_yield_group_104w.update_description(
    "Latest Refused Status Cash loans Contract_type Prior Application's "
    "YIELD_GROUP for the Client over a 104w period."
)
client_count_of_prior_applications_by_prior_application_yield_group_104w.update_description(
    "Count of Prior Applications across different Prior Application "
    "YIELD_GROUP categories for the Client over a 104w period. It provides "
    "a detailed breakdown in the form of a dictionary, where each key "
    "represents a unique YIELD_GROUP. The corresponding value for each key "
    "is the count within that YIELD_GROUP. This distribution offers "
    "insights into the spread across different YIELD_GROUPs for the Client."
)
client_count_of_cash_loans_contract_type_prior_applications_by_prior_application_yield_group_104w.update_description(
    "Count of Cash loans Contract_type Prior Applications across different "
    "Prior Application YIELD_GROUP categories for the Client over a 104w "
    "period. It provides a detailed breakdown in the form of a dictionary, "
    "where each key represents a unique YIELD_GROUP. The corresponding "
    "value for each key is the count within that YIELD_GROUP. This "
    "distribution offers insights into the spread across different "
    "YIELD_GROUPs for the Client."
)
client_avg_of_prior_applications_cnt_payments_104w.update_description(
    "Avg of Prior Applications CNT_PAYMENTs for the Client over a 104w " "period."
)
client_avg_of_approved_status_prior_applications_cnt_payments_104w.update_description(
    "Avg of Approved Status Prior Applications CNT_PAYMENTs for the Client "
    "over a 104w period."
)
client_avg_of_approved_status_prior_applications_amt_annuitys_104w.update_description(
    "Avg of Approved Status Prior Applications AMT_ANNUITYs for the Client "
    "over a 104w period."
)
client_approved_status_prior_applications_amt_credits_by_prior_application_yield_group_104w.update_description(
    "Distribution of the total AMT_CREDITs of Approved Status Prior "
    "Applications, segmented by Prior Application's YIELD_GROUP for the "
    "Client over a 104w period. It provides a detailed breakdown in the "
    "form of a dictionary, where each key represents a unique YIELD_GROUP. "
    "The corresponding value for each key is the cumulative sum of "
    "AMT_CREDITs within that YIELD_GROUP. This distribution offers insights"
    " into the AMT_CREDIT spread across different YIELD_GROUPs for the "
    "Client."
)

In [ ]:

9. Create Window Aggregates from Event Table

Creating Features from the Event Table¶

1. Statistical Aggregations (Avg, Max, Min, Pct)¶

Average (Avg)¶

Percentage (Pct)¶

Maximum (Max)¶

Minimum (Min)¶

2. Features Based on Latest Events and Time Since Last Event¶

Latest Event Values¶

Time Since Latest Event¶

3. Features Based on Inter-Event Time¶

4. Aggregations Across YIELD_GROUP¶

5. Features Derived from Two Other Features¶

Activate catalog¶

Get view from table¶

Create ratio and difference columns¶

Filter View¶

Create Flag column¶

Derive Inter-Event Time columns¶

Do window aggregation from PRIOR_APPLICATIONS¶

Do window aggregation from prior_applications_approved_status_cash_loans_contract_type_view¶

Do window aggregation from prior_applications_approved_status_view¶

Do window aggregation from prior_applications_refused_status_cash_loans_contract_type_view¶

Do window aggregation from prior_applications_revolving_loans_contract_type_view¶

Do window aggregation from prior_applications_cash_loans_contract_type_view¶

Derive Time Since feature from latest timestamp¶

Derive A Ratio from two features¶

Derive Similarity feature across entities¶

Create feature from Inter-Event Time¶

Save features¶

Add description¶

4. Aggregations Across `YIELD_GROUP`¶