https://www.fico.com/blogs/risk-compliance/can-machine-learning-build-a-better-fico-score/
Posted 3/27/18
Comparison of Neural Network to Traditional Logistic based modeling approach:
Our findings may surprise you: We identified very modest predictive lift of less than 2% improvement in KS (the Kolmogorov-Smirnov statistic) from using these cutting edge ML techniques. We examined the scores’ effectiveness across different products (auto, mortgage and bankcard), as well as different lifecycle applications (acquisitions and account management). Consistently, we found that the additional lift offered by the ML-driven algorithms was in that modest 1-2% range.