Closing Session : Portraying the Different Ways Consumers Can Arrive at the Same FICO® Score Value

Presenter Information/ Coauthors Information

Gerald Fahner, FICOFollow

Presentation Type

Oral

Track

Finance/Insurance Application

Abstract

FICO Scores are calculated using many different pieces of consumers’ credit report data, but which can be grouped into five comprehensible categories: payment history, amounts owed, length of credit history, new credit, and credit mix. Consumers can gain qualitative insights into what's helping or straining their score by reviewing their personal ratings (graded from poor to exceptional) for each category available at myFICO.com.

Our latest research enhances score explanations by computing five credit category scores that add up to the FICO® Score. In this presentation we will share our innovative explainable machine learning approach to generate category scores, which has been tailored to the tree-segmented Generalized Additive Model structure of the FICO® Score. We will share new insights from analyzing the category score distribution of the US scorable population. Category score-based clustering and profiling of consumers sharing the same FICO® Score unveils the behavioral variety that can exist at a given score value. Besides having potential applications in consumer credit education, we also see potential value for category scores to inform lender decision making.

Start Date

2-7-2025 4:00 PM

End Date

2-7-2025 5:00 PM

This document is currently not available here.

Share

COinS
 
Feb 7th, 4:00 PM Feb 7th, 5:00 PM

Closing Session : Portraying the Different Ways Consumers Can Arrive at the Same FICO® Score Value

Volstorff B

FICO Scores are calculated using many different pieces of consumers’ credit report data, but which can be grouped into five comprehensible categories: payment history, amounts owed, length of credit history, new credit, and credit mix. Consumers can gain qualitative insights into what's helping or straining their score by reviewing their personal ratings (graded from poor to exceptional) for each category available at myFICO.com.

Our latest research enhances score explanations by computing five credit category scores that add up to the FICO® Score. In this presentation we will share our innovative explainable machine learning approach to generate category scores, which has been tailored to the tree-segmented Generalized Additive Model structure of the FICO® Score. We will share new insights from analyzing the category score distribution of the US scorable population. Category score-based clustering and profiling of consumers sharing the same FICO® Score unveils the behavioral variety that can exist at a given score value. Besides having potential applications in consumer credit education, we also see potential value for category scores to inform lender decision making.