Scorecard

Credit scoring

• Efficient assessment of customer credit risk
• Quick approval or rejection of loan applications based on objective scoring criteria
• Ensure a high level of security
• Reduce the number of unpaid debts

The Scorecard is commonly applied in the financial sector, however it is also efficient in any other type of business, where we need to decide whether to grant a merchant credit on a regular basis.

Debt collection (recovery) optimization

• Identification of debts likely to become bad debts
• Identification of bad debts that can still be recovered profitably
• Identification of self-cured debts that do not require additional action
• Identification of the most effective collection approach for each debtor
• Identification of debts highly unlikely to be recovered and suitable for sale

Fraud detection

• Identify transactions, claims, and activities with a high probability of fraud
• Detect suspicious behavior and potential abuse early
• Reduce financial losses through proactive fraud prevention
• Enable faster response to high-risk events
• Build customer trust with strong security standards

Marketing scoring

• Prevent customer churn
• Enable effective up-selling opportunities
• Support targeted cross-selling strategies

The Scorecard is an effective tool for identifying groups of customers (or other entities) that require specific actions, such as targeted product offers.

Medical scoring

• Patient health status modeling: sick vs. healthy
• Prediction of the risk of post-procedure complications
• Analysis and prediction of selected patient medical characteristics

KEY FEATURES:

Select important features from the data

• Exclude unimportant or redundant variables
• Create a variable ranking of overall predictive powerj
• Identify redundancy among variablesi

Find subgroups representing high risk

• Search rules enabling the identification of high-risk subgroups
• Use advanced data mining algorithm to identify rules

Build risk profiles for each characteristic

• Shape the awareness of the impact of each variable on risk
• Manage “dirty data” – outliers, missing data, and nonlinearities

Create industry standard in-house scorecards

• Quickly develop in-house scorecards
• Use advanced predictive analytics to build the model
• Manage rejected applications

Validate and monitor models

• Check the predictive power of the model
• Find the optimal pricing strategy
• Examine population stability

MODULES:

Data preparation

• Feature Selection
• Interactions and Rules
• Attribute Building

Modeling

• Scorecard Preparation
• Survival
• Rejected Inferences

Evaluation and calibration

• Model Evaluation
• Cut-off Point Selection
• Score Cases
• Calibration Tests

Monitoring

• Population Stability
• Vintage Analysis
• Migration Matrices