Credit Scoring for SAS® Enterprise Miner™
Flexible data preparation and management
- SAS data access, integration and management capabilities make it easier to prepare data across disparate systems and sources for analysis.
- Credit risk models generalize well and produce superior outcomes.
- Comprehensive variable selection techniques help lead to better credit risk modeling.
SAS Enterprise Miner: Award-winning predictive analytics
- The most comprehensive set of advanced predictive and descriptive modeling algorithms, including scorecard, decision trees, neural networks, logistic regression, etc.
- Assesses scorecard quality to determine the best course of action.
Patented optimal rigorous binning method
- The SAS patented optimal rigorous binning method yields true optimal bins based on constraints defined by the user.
Data partition node
- Partitions data into training, validation and/or test data sets.
- Data sets based on simple random, cluster or stratified random sample.
Interactive grouping node
- Credit scorecard and screening and binning of univariate characteristics.
- Purposeful censoring of data and assessments on the strength of each characteristic individually as a predictor of performance.
- Use of predefined groupings.
- Prebinning through quantile or bucket methods.
- Interval binning through optimal criterion, quantile, monotonic event rate and constrained optimal methods.
- Interactive bin changes.
- Supports interval targets.
- Variable selection through Gini statistic or information value.
- Response rate plots.
- Gini statistic and information value plots.
- Output of variable mappings.
- Fits a logistic regression model and computes the scorecard points for each attribute.
- Choose either the WOE variables or the group variables that are exported by the Interactive Grouping node as inputs for the logistic regression model.
- Scorecard points of each attribute based on the coefficients of the logistic regression model.
- Manually assign scorecard points to attributes.
- Scaling of scorecard points, controlled by three Scaling Options properties.
- Intercept-based scorecard.
- Reverse scorecard.
- Use of indeterminate values in gains charts.
- Adverse characteristic analysis through neutral score or weighted average score methods.
- Inputs for trade-off plots.
- Score distribution charts.
- Strategy curve.
- Event frequency chart.
- Scorecard strength chart.
- Selection methods include backward, forward and stepwise.
- Regression criteria include AIC, SBC, validation error, validation misclassification rate, cross-validation error, and cross-validation misclassification rate.
Reject inference node
- Provides remedies for the sample/population data selection bias through an augmented data set that represents the "through-the-door" population; this serves as training data set for a second scorecard model.
- Inference methods include fuzzy, hard cutoff and parceling.
- Classification charts of actual versus inferred.
- Distribution plots of actual versus inferred.
- Summary statistics of actual versus inferred.
Seamless integration into SAS Credit Scoring
- SAS Credit Scoring is a comprehensive solution for developing, deploying and managing scorecards for operational and regulatory compliance that includes data management at the front end and model management at the back end.