Scorecard is any Risk Model belonging to the class of typically (but not necessarily) simple quantitative models that are used to construct numerical estimates of risk (Scores) on the basis of a defined list of input data.
A scorecard model contains a set of input fields (values), also named characteristics which are used to infer (predict) one (or more) output values (scores). This inference can be seen as an assessment about the likelihood of a future event affecting a customer, a transaction or any other concrete entity or system (e.g. the probability of default or failure). Characteristics are encoded in attributes with specific partial scores associated with them. Partial scores are then summed up so that an overall score can be obtained for the target value. The score summarizes the influence of the input attributes on the outcome. It is readily available for inspection. Input attributes can be mapped to reason codes which provide explanations for any specific score calculation.
Scorecards are popular in various industries but in particular in the Financial Industry. They are preferred for their interpretability, relative ease of implementation which allows purely statistical, expert-based or combined (hybrid) approaches. Expert Scorecards are particularly useful in cases where statistical estimation is hindered by lack of data.
- Credit Scoring using Credit Scorecard
- Fraud Risk
- Customer Attrition (or Churn, or Retention)
- Customer Marketing Response
- Product Usage