Angoss ScorecardBUILDER

Build Better Scorecards, Faster Increase Efficiency

Develop Strategies

Minimize Bias

Select Best Variables

Optimize Binning

Introduction Build Better Scorecards, Faster Angoss ScorecardBUILDER

Credit scoring and automated rule-based decisioning are the most important tools used by financial services and credit lending organizations for credit lifecycle management. They are commonly used in consumer lending for products such as credit cards, personal unsecured loans, lines of credit and residential mortgages.

Best practices in the financial services have moved toward the use of proprietary credit scores and strategies Increase Efficiency created using data mining and predictive analytics. These allow financial services organizations to augment and extend generic scores and scorecard based systems in order to make more differentiated decisions at the individual borrower level. Predictive analytics capabilities and the ability to act upon their insights and best actions throughout the credit decisioning lifecycle is recognized as a strategic competitive advantage resulting in improved profitability and market share growth—particularly for customer acquisition and retention strategies.

Develop Strategies Select with theBest use ofVariables This eBook will explore 5 key areas where faster, better scorecard development can be enabled predictive analytics.

Increase Efficiency

Select Best Minimize Variables

Optimize BiasBinning

Minimize Bias

Develop StrategiesOptimize

ScorecardBUILDER

Binning

Increase Efficiency

Select Best Variables

Optimize Binning

Minimize Bias

Increase Efficiency Develop scorecards more efficiently with automated workflow. An automated visual canvas for building and displaying scorecard workflows allows for the refreshing and reusing of scorecard development workflows in minutes - eliminating the need to write code, increasing efficiency. With an automated workflow, process nodes can simply be dragged onto the canvas where they would be connected to form a workflow and run individually or in batch sequentially with the ability to add new data or refresh with the click of a button. A workflow provides instant visual documentation of the analytical tasks required in the credit risk scorecard development process.

Develop Strategies

ScorecardBUILDER

Step 1 Increase Efficiency with Automated Workflow: A visual canvas enables the creation of a scorecard workflow in minutes

Increase Efficiency

Select Best Variables

Optimize Binning

Minimize Bias

Select Best Variables Select the most predictive variables for best scorecard performance. Credit data sets can have thousands of variables; however, in most cases an effective scorecard requires only 10-20 of the most predictive variables. The efficient analysis of thousands of candidate predictors for easy selection of the most predictive, influential variables is needed. A variety of methods can be used to enable users to see and understand relationships and behaviours - quickly providing accurate assessments to support variable reduction including visual exploration of variables with characteristic analysis, Decision Trees, and various charts and tables. Measures of Predictive Power allow users to effortlessly sift through variables and in a matter of minutes select the most predictive variables to use in scorecard development. Candidate predictors can be sorted using any of the predictive measures with the click of a button, greatly reducing the overall time required in scorecard variable selection.

Develop Strategies

ScorecardBUILDER

Step 2 Measures of Predictive Power: A variety of methods can be used to enable users to see and understand relationships and behaviours.

Payment history

Types of credit used

Debt burden Own home or rent Recent searches for credit

Increase Efficiency

Select Best Variables

Optimize Binning

Minimize Bias

ScorecardBUILDER

Develop Strategies

Optimize Binning Automate coarse classing with flexible WOE optimizer. Credit scorecards assign points to a range of values. As a result, coarse classing (binning) is necessary in order to aggregate data into stable and statistically significant ranges and develop meaningful target variable trends across these ranges. This is a traditionally tedious manual task, which can be automated with a flexible and intelligent Weight of Evidence Optimizer (WOE) – reducing time spent on this manual task by up to 50%. Automating the coarse classing process speeds up the time-consuming task of optimizing predictive variable bins which is necessary to ensure monotonic trends, no null values, and equal sized bin ranges, which in turn improves the stability and accuracy of the scorecard.

Step 3 Weight of Evidence Optimizer (WOE): Reduce time spent on this manual task by up to 50%

Not Family 2.87765

Spouse WOE

Other WOE Not Family WOE Sibling WOE

Other 1.99825 Spouse 1.17765

Sibling 1.50978

Increase Efficiency

Select Best Variables

Optimize Binning

Minimize Bias

Minimize Bias Remove selection bias to produce the most effective application scorecards. Traditional application scorecards make it difficult for financial institutions to capture all “good” customers. Inevitably, some customers that are accepted will default on a loan, or payment, while some who are rejected might be creditworthy and profitable. Reject Inference improves the quality of application scorecards by correcting and minimizing the selection bias and preventing model overfit. Reject inference methods including Proportional Assignment, Hard Cutoff, Parceling, and Fuzzy Augmentation ensure more accurate and realistic performance of credit scoring models producing scorecards which predict the behavior of the total population. Ultimately, this enables lending and credit organizations to maximize their business growth by optimizing the risks and rewards for individual lending decisions.

Develop Strategies

ScorecardBUILDER

Step 4 Reject Inference: Improves the quality of application scorecards by correcting and minimizing the selection bias and preventing model

Increase Efficiency

Select Best Variables

Optimize Binning

Minimize Bias

ScorecardBUILDER

Develop Strategies

Develop & Deploy Strategies Seamlessly develop, and deploy customer strategies using scorecard results. Strategy design, development and deployment tools provide users with the ability to combine scorecards with user-defined business rules and key performance indicators (KPIs) in order to produce highly targeted lists to which treatments such as credit limit, collections and marketing campaign activities can be applied.

Step 5 Predictive strategies: Combine scorecards with business rules and key performance indicators (KPIs) in order to produce highly targeted lists.

Predictive strategies resulting from the combination of scorecard models and business rules, deliver significantly higher impact than non-predictive business rules and result in a highly automated and optimized decisioning environment that yields significant user and organizational benefits at each stage of the credit lifecycle process. Customer

Angoss ScorecardBUILDER Enabling faster, better scorecard development as a seamless step in the predictive analytics workflows.

Develop Strategies Develop, deploy and monitor customer strategies in ScorecardBUILDER with Angoss patented Strategy Trees – a unique approach for building and deploying

Minimize Bias ScorecardBUILDER provides 4 reject inference methods to improve the quality of application scorecards by correcting and minimizing the selection bias and preventing model overfit.

Optimize Binning ScorecardBUILDER automates this traditionally manual task, with a flexible and intelligent Weight of Evidence Optimizer (WOE) – reducing time spent on this by up to 50%.

Select Best Variables ScorecardBUILDER enables the efficient analysis of thousands of candidate predictors for easy selection of the most predictive, influential variables.

Increase Efficiency ScorecardBUILDER provides a powerful automated canvas for building, refreshing, and reusing scorecard development workflows.

Angoss ScorecardBUILDER Angoss ScorecardBUILDER is a robust, end-to-end, scorecard building product which enables credit risk analysts to build and deploy precise, industry-compliant scorecards faster by automating time-consuming, manual steps. ScorecardBUILDER provides analysts with the ability to create high quality credit scorecards as a seamless step in their analytics workflows, from data profiling and segmentation through predictive modeling and scorecard building to strategy design and deployment. To learn more visit Angoss.com.

Quick Tour View a quick tour of Angoss ScorecardBUILDER to learn how to build better scorecards, faster.

Watch Video

Data Sheet Learn more about Angoss ScorecardBUILDER features & benefits.

Download

Webcast Learn more about how Angoss ScorecardBUILDER automates the most time consuming scorecard development steps resulting in more than 50% time savings.

Watch Video

About Angoss About Angoss Software. Angoss is a global leader in delivering predictive analytics to businesses looking to improve performance across risk, marketing and sales. With a suite of desktop, client-server and big data analytics software products and cloud solutions, Angoss delivers powerful approaches to turn information into actionable business decisions and competitive advantage. Angoss software products and cloud solutions are user-friendly and agile, making predictive analytics accessible and easy to use. Headquartered in Toronto, Canada, Angoss has offices in the United States and United Kingdom. For more information, visit www.angoss.com.