Here is a list of potential features for the package. The priority would be to have one feature for each subsection from the preprocessing section and the model development section. Each subsection could be independant sets of methods.

Preprocessing

Data structure

• Integrated data structure (defined target and features, already binned variable)

1. Defined target and feature vectors
2. date
3. integrated train and test set, linked with date
4. integrated segmentation

Default inference

• Default inference for declined loan demands
• A different inference type

Discretization

• (automated) WOE binning (variable discretisation)
• Other Binning techniques

Model development

Model design tools

• Automated segmentation research
• Automated variable selection

Model types

• Scorecard development
• Shadow rating (for low default portfolio)

Calibration grid and rating tools

• Probability of defaut given score
• Calibration tools
• Caligration given certain cutoff points
• Recalibration tools

Regulatory backtesting tools

Discrimination power

• Vasicek and wilcoxon
• Auroc
• Crystal ball (for shadow rating)

Population stability

• Population stability index