He showed that machine learning models using alternative data can score 70-80% of the previously unscoreable, though with higher model risk.
by L. C. Thomas Hot
The techniques for handling missing data, outliers, and imbalanced datasets are timeless. 5. Conclusion credit scoring and its applications by l c thomas hot
While initially designed for personal loans and mortgages, Thomas, Edelman, and Crook highlighted the versatility of these techniques. The applications extend far beyond traditional banking: He showed that machine learning models using alternative
To address this, the field is embracing a wealth of new data points: credit scoring and its applications by l c thomas hot
The methodologies for validating, monitoring, and updating scorecards remain standard industry practice.
While primarily used in banking and finance, the techniques described by Thomas have been adapted for several non-traditional fields: The University of Texas at Austin Direct Marketing