with a better browsing experience; allow us to assess, monitor, and improve the website’s
performance; and enable our partners to advertise to you. You may disable the cookies by changing
the settings in your browser, and you may tell us not to share your cookie data with third parties.
Generally available consumer credit scoring models can provide great value to lenders in evaluating the risk of loan applicants. These models are designed to distill the predictive power of a large number of consumer factors into a single, easily understood score. While the credit scoring models work well in estimating default likelihood over time, all models eventually may need to be evaluated to determine if an update to the existing model is needed, or whether the existing model should be replaced by a new credit score model. However, these evaluations can be challenging. For example, comparing the performance of the new credit scoring model to an existing model requires meticulous attention to detail as subtle statistical quirks can easily bias results.