Re: How to model/test some financial data
- From: "Phil Sherrod" <PhilSherrod@xxxxxxxxxxxxxxxxxxxxx>
- Date: Sat, 1 Mar 2008 02:07:00 GMT
On 29-Feb-2008, ourproperty@xxxxxxxxx wrote:
P1 - Borrowers that are members of groups are more likely to get
funded that those that are not.
P2 - Borrowers that are members of groups are more likely to secure a
lower interest rate than those that are not.
P3 - Borrowers that are members of groups are more likely to secure a
greater rate saving than those that are not.
When you do your tests comparing "groups", you need to consider other factors
that may be correlated with membership in the groups. For example, if group A
has 10% college graduates and has an average loan default rate of 20%, and
group B has 60% college graduates and a loan default rate of 2%, you can't say
that the outcomes are determined by simply being an A or B type person. You
need to balance the other factors such as income, education, age, marital
status, net worth and loan history when making the comparisons.
--
Phil Sherrod
(PhilSherrod 'at' comcast.net)
http://www.dtreg.com (Decision trees, Neural networks, SVM and Genetic
modeling)
http://www.nlreg.com (Nonlinear Regression)
.
- Follow-Ups:
- Re: How to model/test some financial data
- From: Gleanna
- Re: How to model/test some financial data
- Next by Date: Re: How to model/test some financial data
- Next by thread: Re: How to model/test some financial data
- Index(es):
Relevant Pages
|
