Extending the Credit Offers decision strategy with Credit Score

The Credit Scores flow assesses the customer's creditworthiness using internal scoring models and external credit bureau data.

These scores are crucial for evaluating the risk associated with lending to a particular customer. By combining internal scores with credit bureau scores, the strategy provides a comprehensive risk profile.

Start by inspecting the sample application's customer data model. Then, extend the strategy with a second flow to add more intelligence by utilizing external credit bureau data. Develop a scorecard model to use internal data to assess credit risk, and combine the external and internal model scores. Based on that final credit risk score, you create a decision table to determine risk-based pricing tiers.

You can access data from Prediction Studio. The SimulationSetDDS data set contains simulated customer data. To familiarize yourself with this test data, browse the data set with a customer ID as the key.

The customer data model is a nested data structure which includes customer profile data combined with additional credit related data on the CreditResponse sub-page. Customer profile data contains information such as Region Code, Date of Birth, Marital status, and so on.

The CreditResponse page of the customer data includes credit liabilities, credit scores, and a credit summary for each customer.

For example, this customer has three credit liabilities, and the output credit score is 650:

This CreditResponse data is coming from an external credit bureau data. Customers data is made availablepage before the strategy is executed.

In a real-life implementation, the Customer data model can contain many additional data sources. Typically, a data flow is used to prepare all the necessary data to make a decision.

To integrate the Equifax credit bureau data into the Credit Offers decision strategy, you use a pre-build sub-strategy.

The EquifaxResponse strategy processes the customer information requested from the Credit Bureau. By default, this data includes the credit score, total liabilities, total delinquency, and AT57S value of the customer.

The EquifaxResponse strategy uses the CreditResponse embedded strategy component to aggregate credit liabilities and delinquencies and output the total monthly payments and the number of missed payments.

The CreditResponse is an iteration component. The Iteration component in a decision strategy is used to loop through a list of propositions, pages, or values. It allows you to apply a set of actions or decisions to each item in the list. This is particularly useful when you want to evaluate or manipulate multiple items in a decision strategy.

In this case, the component iterates over the liability pages to aggregate credit liabilities and delinquencies.

The top-level Primary is the customer page, and .CREDIT_RESPONSE and .CREDIT_LIABILITY reflect the hierarchy of the customer data model.

Each entity is accessible in the CreditLiability property, for example, CreditLiability.MonthlyPaymentAmount.

In each iteration, the embedded strategy component sets the value for the TotalDelinquency property by summing up the LateCount values.

The Expression that sets the TotalLiabilities property uses the IsClosedIndicator property to exclude closed credit liabilities.

The embedded strategy component outputs a single aggregated value for the TotalLiabilities and TotalDelinquency properties.

The CreditSummary embedded strategy component outputs the total past due amount of open trades verified in the past 12 months as the AT57S property that the credit bureau provides.

The Filter component selects the AT57S code, and the Set property component assigns the value to the property.

Finally, the strategy reads the Score value directly from the Data Set, and references the values for the TotalLiabilities, TotalDelinquency, and AT57S properties.

When we run the strategy, these properties are computed based on the credit bureau data.

The business wants to use a scorecard to assess credit risk based on factors such as marital status, employment status, accommodation type, and whether the customer has insurance.

To meet this requirement, you configure a Scorecard model component in the Credit Offers decision strategy.

A scorecard assigns a score to a value or a range of values for a particular predictor expression.

To assign a score for employment duration, you use a formula that calculates the time difference between the value of the LastEmploymentStartDate property and today.

Then you define the score ranges. A longer employment duration gives a higher score.

You use the MaritalStatus property to give married and widowed customers a higher score than divorced and single customers.

The combiner function determines how the scores for the properties are processed. In this case, you sum up the scores.

When all relevant properties are added, you define score ranges and map them to the RED, YELLOW, and GREEN segments to create a comprehensive risk profile.

The business wants to use the newly built scorecard in combination with the Equifax Credit Bureau data to set a score segment.

To produce a final score segment, you use a decision table to combine the scorecard model results with Equifax data.

In the decision table, define conditions that combine the internal and external scores to produce a final score segment. The internal score, .Score, is the score determined by the Credit Application scorecard and the external score, .CBScore, is the score value returned by Equifax Credit Bureau.

This categorizes customers into different risk segments based on combined scores.

The business wants to implement a pricing model to tailor offers to the customer's risk profile. To meet this requirement, create a decision table to determine risk-based pricing tiers.

You must configure the table to set pricing properties based on the combined risk assessments from the scorecard and Equifax data and the loan offer. The customer Segment, is the segmentation result of the Credit Application scorecard.

Next, you configure properties for the calculations that are necessary for determining the final credit offer, such as maximum payment, interest rate, and offer amount. To meet this requirement, use a set property component.

Configure the Set Property component to include the properties for determining the final credit offer.

Next, implement a filter component to ensure that only offers meeting minimum criteria are selected.

In this case, you check if the offer equal to or more than the minimum amount, a property set on the individual loan proposition.

Finally, to streamline the process, the business wants to distinguish between credit score outputs and credit offer outputs within the decision strategy. For this requirement, create a new output flow for credit scores using a Set Property component.

Configure the component to output relevant credit score information separately from credit offers, based on the final score segment.

To test your work, you run the strategy to ensure that the credit score and credit offer outputs are correctly processed.

Note the two pages for the loan offer and the scoring segment: