Systems and methods for determining thin-file records and determining thin-file risk levels
In some embodiments, systems and methods are disclosed for generating filters to determine whether a consumer is likely to have a scoreable credit record based on non-credit data, and to determine a potential risk level associated with an unscoreable credit record based on non-credit data. Existing scoreable and unscoreable records are compared to determine factors correlated with having an unscoreable record, and a multi-level filter is developed. Unscoreable records having at least one entry are compared to determine whether they are “good” or “bad” risks, factors correlated with either condition are determined, and a filter is developed. The filters can be applied to records comprising demographic data to determine consumers that are likely to have unscoreable records but represent good risks.
1. A computer system, comprising:
a thin-file propensity model stored in non-transitory computer storage; and
one or more computing devices comprising one or more hardware processors configured to:
access the thin-file propensity model from the computer storage, wherein generation of the thin-file propensity model involved an identification of one or more first demographic characteristics correlated to one or more records that include less than a predetermined amount of credit data,
wherein the identification was based at least on a likelihood that other records also associated with the first demographic characteristics include less than the predetermined amount of credit data;
apply the thin-file propensity model to a first record to:
determine that one or more demographic characteristics associated with a first record match at least one of the first demographic characteristics; and
determine that the first record likely includes less than the predetermined amount of credit data; and
determine a risk score associated with the first record using at least one demographic characteristic associated with the first record, wherein the risk score determination comprises applying a risk score model to the first record using the at least one demographic characteristic as an input to the risk score model.
2. The computing system of claim 1 ,
wherein the first demographic characteristics comprise one or more of residence address data, age data, household data, marital status data, delinquent data for consumers in a geographic area, and data related to length of residency.
3. The computing system of claim 1 ,
wherein one or more of the first demographic characteristics are the same as the at least one demographic characteristic associated with the first record.
4. The computing system of claim 1 ,
wherein each of the first demographic characteristics is different than the at least one demographic characteristic associated with the first record.
5. The computing system of claim 1 ,
wherein the at least one demographic characteristic associated with the first record comprises one or more of residence address data, age data, household data, marital status data, delinquent data for consumers in a geographic area, and data related to length of residency.
6. A method of identifying a level of risk associated with a thin-file credit record, comprising:
accessing a thin-file propensity model from non-transitory computer storage, wherein generation of the thin-file propensity model involved an identification of one or more first demographic characteristics correlated to one or more records that include less than a predetermined amount of credit data,
wherein the identification was based at least on a likelihood that other records also associated with the first demographic characteristics include less than the predetermined amount of credit data;
using one or more hardware processors, applying the thin-file propensity model to a first record to:
determine that one or more demographic characteristics associated with the first record match at least one of the first demographic characteristics; and
determine that the first record likely includes less than the predetermined amount of credit data; and
using one or more hardware processors, determining a risk score associated with the first record using at least one demographic characteristic associated with the first record, wherein determining the risk score comprises applying a risk score model to the first record using the at least one demographic characteristic as an input to the risk score model.
7. The method of claim 6 ,
wherein the at least one demographic characteristic associated with the first record comprises one or more of residence address data, age data, household data, marital status data, delinquent data for consumers in a geographic area, and data related to length of residency.
8. The method of claim 6 ,
wherein each of the first demographic characteristics is different than the at least one demographic characteristic associated with the first record.
9. The method of claim 6 ,
wherein one or more the first demographic characteristics are the same as the at least one demographic characteristic associated with the first record.
10. The method of claim 6 ,
wherein the first demographic characteristics comprise one or more of residence address data, age data, household data, marital status data, delinquent data for consumers in a geographic area, and data related to length of residency.
11. Non-transitory computer storage comprising instructions which, when executed, direct a computer system to perform a method comprising:
accessing a thin-file propensity model from non-transitory computer storage, wherein generation of the thin-file propensity model involved an identification of one or more first demographic characteristics correlated to one or more records that include less than a predetermined amount of credit data,
wherein the identification was based at least on a likelihood that other records also associated with the first demographic characteristics include less than the predetermined amount of credit data;
using one or more hardware processors, applying the thin-file propensity model to a first record to:
determine that one or more demographic characteristics associated with a first record match at least one of the first demographic characteristics; and
determine that the first record likely includes less than the predetermined amount of credit data; and
using one or more hardware processors, determining a risk score associated with the first record using at least one demographic characteristic associated with the first record, wherein determining the risk score comprises applying a risk score model to the first record using the at least one demographic characteristic as an input to the risk score model.
12. The non-transitory computer storage of claim 11 ,
wherein the first demographic characteristics comprise one or more of residence address data, age data, household data, marital status data, delinquent data for consumers in a geographic area, and data related to length of residency.
13. The non-transitory computer storage of claim 11 ,
wherein each of the first demographic characteristics is different than the at least one demographic characteristic associated with the first record.
14. The non-transitory computer storage of claim 11 ,
wherein one or more of the first demographic characteristics are the same as the at least one demographic characteristic associated with the first record.
15. The non-transitory computer storage of claim 11 ,
wherein the at least one demographic characteristic associated with the first record comprises one or more of residence address data, age data, household data, marital status data, delinquent data for consumers in a geographic area, and data related to length of residency.