Algorithm for explaining credit scores
An exemplary Web-based score explanation service typically requires only the credit bureau identifier, credit score, and up to four reason codes as input. The invention herein discloses an algorithm that is used to provide an explanation of the primary factors influencing the score, where a rich data feed is provided to the facility implementing the algorithm
1. A computer-implemented method for explaining credit scores, the method comprising:
providing a Web site on a web server that contains informative resources, the Web site:
providing consumers access to the informative resources via a network connected to the web server, the informative resources comprising information representing practices of the collecting, storing, reporting, and evaluating of consumer credit data;
receiving consumer credit scores and reason codes from individual consumers or third parties, in interactive or batch modes;
providing an explanation report to the individual consumers based upon the consumer credit scores associated with each individual consumer;
defining a credit score as a function of prediction characteristics;
defining a surrogate set of characteristics as representing areas for improvement;
developing, by at lest one data processor, by a server, a surrogate score to approximate a real credit score, using said real credit score as a performance dependent variable and using said surrogate characteristics as predictors; and
defining a potential improvement metric for each area for improvement.
2. The method of claim 1 , wherein the Web site further creates an ordered areas for improvement table.
3. The method of claim 1 , wherein the Web site further uses said surrogate characteristics as said areas for improvement.
4. The method of claim 1 , wherein the Web site further defines a set of surrogate characteristics from a standard categorization of original prediction characteristics.
5. The method of claim 4 , wherein the Web site further develops surrogate characteristics as models developed using said categorized prediction characteristics and said credit score as said performance variable.
6. The method of claim 5 , wherein the Web site further creates said areas for improvement.
7. The method of claim 1 , wherein said defining a credit score step is defined as:
Score=ζ( x 1 ,x 2 , . . . ,x c ),
where
x j =Prediction Characteristic j.
8. The method of claim 1 , wherein said developing a surrogate score step further comprises the step of: using said areas for improvement prediction variables to develop a surrogate score of the form
Ψ( z 1 ,z 2 , . . . ,z p )
which is developed using z 1 , z 2 , . . . , z p as the prediction characteristics and,
y =ζ(χ 1 ,χ 2 , . . . ,χ c )
as said performance variable; and
wherein said defining a potential improvement metric step is defined as:
I
k
=
100
⨯
max
z
k
{
ψ
(
z
1
*
(
z
k
)
,
…
,
z
k
,
…
,
z
p
*
(
z
k
)
)
-
ψ
(
z
1
*
,
…
,
z
k
*
,
…
,
z
p
*
)
}
Score
*
=
Maximum
possible
percent
improvement
in
score
for
Area
of
Improvement
k
.
where z 1 *(z k )=z 1 * unless z 1 * cannot coexist with z k , wherein in that case, z 1 *(z k )=E[z 1 |z k ], or some other value of z 1 that can coexist with z k .
9. The method of claim 1 , wherein the Web site further:
associates a set of reason codes with a score returned from a credit bureau;
associates a score difference with each reason code;
provides a data feed which includes score differences associated with said reason codes; and
converts score differences to percentages.
10. The method of claim 1 , wherein the Web site further compares a current score to a maximum score that can be obtained by varying said prediction characteristic.
11. A credit score explanation system comprising:
a Web site running on a server that contains informative resources, said Web site comprising any of for-pay services and extranet/internet functions;
said Web site offering any of consumers and said third parties access to information contained in said informative resources, the resources being both general and personal, about practices of any of collection, storing, reporting, and evaluating consumer credit data; and
said Web site for accepting consumer credit scores and reason codes from any of individual consumers or third parties, in interactive or batch mode, for providing an explanation report to said individual consumers based upon the individual consumers' credit scores, for defining a credit score as a function of prediction characteristics, for defining a surrogate set of characteristics as representing areas for improvement, and for developing a surrogate score to approximate a real credit score, using said real credit score as a performance dependent variable and using said surrogate characteristics as predictors, and for defining a potential improvement metric for each area for improvement.
12. The system of claim 11 , wherein the Web site further creates an ordered areas for improvement table.
13. The system of claim, 11 , wherein the Web site further uses said surrogate characteristics as said areas for improvement.
14. The system of claim 11 , wherein the Web site further defines a set of surrogate characteristics from a standard categorization of original c prediction characteristics.
15. The system apparatus of claim 14 , wherein the Web site further develops surrogate characteristics as models developed using said categorized prediction characteristics and said credit score as said performance variable.
16. The system of claim 15 , wherein the Web site further creates said areas for improvement.
17. The system of claim 11 , wherein defining a credit score comprises:
Score=ζ( x 1 ,x 2 , . . . ,x c ),
where
x j =Prediction Characteristic j.
18. The system of claim 11 , wherein said developing a surrogate score step further comprises: using said areas for improvement prediction variables to develop a surrogate score of the form
Ψ( z 1 ,z 2 , . . . ,z p ),
which is developed using z 1 , z 2 , . . . , z p as the prediction characteristics and,
y =ζ(χ 1 ,χ 2 , . . . ,χ c )
as said performance variable.
19. The system of claim 11 , wherein defining a potential improvement metric step is defined as:
I
k
=
100
⨯
max
z
k
{
ψ
(
z
1
*
(
z
k
)
,
…
,
z
k
,
…
,
z
p
*
(
z
k
)
)
-
ψ
(
z
1
*
,
…
,
z
k
*
,
…
,
z
p
*
)
}
Score
*
=
Maximum
possible
percent
improvement
in
score
for
Area
of
Improvement
k
.
where z 1 *(z k )=z 1 * unless z 1 * cannot coexist with z k , wherein in that case, z 1 *(z k )=E[z 1 |z k ], or some other value of z 1 that can coexist with z k .
20. The system of claim 11 , wherein the Web site is further adapted for:
associating a set of reason codes with a score returned from a credit bureau;
associating a score difference with each reason code;
providing a data feed which includes score differences associated with said reason codes; and
converting score differences to percentages.
21. The system of claim 11 , wherein the Web site further compares a current score to a maximum score that can be obtained by varying said prediction characteristic.
22. An article of manufacture comprising:
computer executable instructions stored on non-transitory computer readable media, which, when executed by a computer, causes the computer to perform operations comprising:
providing a Web site on a web server that contains informative resources, the Web site:
providing consumers access to the informative resources via a network connected to the web server, the informative resources comprising information representing practices of the collecting, storing, reporting, and evaluating of consumer credit data;
receiving consumer credit scores and reason codes from individual consumers or third parties, in interactive or batch modes;
providing an explanation report to the individual consumers based upon the consumer credit scores associated with each individual consumer;
defining a credit score as a function of prediction characteristics;
defining a surrogate set of characteristics as representing areas for improvement;
developing a surrogate score to approximate a real credit score, using said real credit score as a performance dependent variable and using said surrogate characteristics as predictors; and
defining a potential improvement metric for each area for improvement.
23. The article of claim 22 , wherein developing a surrogate score step further comprises:
using said areas for improvement prediction variables to develop a surrogate score of the form
Ψ( z 1 ,z 2 , . . . ,z p ),
which is developed using z 1 , z 2 , . . . , z p as the prediction characteristics and,
as said performance variable; and
wherein said defining a potential improvement metric step is defined as:
I
k
=
100
×
max
z
k
{
Ψ
(
z
1
*
(
z
k
)
,
…
,
z
k
,
…
,
z
p
*
(
z
k
)
)
-
Ψ
(
z
1
*
,
…
,
z
k
*
,
…
z
p
*
)
}
Score
*
=
Maximum
possible
percent
improvement
in
score
for
Area
of
Improvement
k
.
where z 1 *(z k )=z 1 * unless z 1 * cannot coexist with z k , wherein in that case, z 1 *(z k )=E[z 1 |z k ], or some other value of z 1 that can coexist with z k .
24. The article of claim 22 , wherein the Web site further:
associates a set of reason codes with a score returned from a credit bureau;
associates a score difference with each reason code;
provides a data feed which includes score differences associated with said reason codes; and
converts score differences to percentages.