2016-03-03 Meeting notes
- Pawel Garbacz
- Lynn Calahan [X] (Unlicensed)
- Dennis Wisnosky
- Michael
Date
Attendees
Agenda
FIBO Loans FCT Meeting Agenda_030316.doc
1) Use Case reminder.
2) Where we are on our road map.
3) Open Action Items
4 ) JIRA Issues Review
5) Todays content discussion.
Loan Contract Data Decompostion v151014a.vsd
Loan Contract Data Decompostion v151105.vsd
Please note the updated Loan contract diagram V has provided above.
UML
Spreadsheet LoansAnnotations-Integrated-2016-03-03.xlsx
Protégé loanCore 2016-03-03.rdf
6) For next week.
Proceedings
Minutes:
FIBO Loans FCT Meeting Agenda_030316_minutes.doc
The updated Annotations spreadsheet:
LoansAnnotations-Integrated-2016-03-03_team comments.xlsx
New person: Carolina Cabanas-Leon IOFD/DOFI
hasCollateral: definition could have “security agreement”, but that wsa not in the notes. OK for now.
Went through changes form last week, mostly minor wording or spelling changes.
New subject: credit.
Reviewed Viesturs Lenss’s summary of loan contract information.
Lenss: the appraisal is assessment of collateral value. Can make appraisal attribute of real property, also could be part of credit risk assessment.
LC/MU: can be connected to both, the great thing about semantics.
LC: two sense of credit, bit C the whole assessment, vs. little ‘c’ the credit score part.
JeffB: credit risk very braod, consumer, small business, corporate. Different credit risk based on type of loan to what kind of borrower. CRdit score is just one imple of a risk model.
LC: yes. The credit reporting agencies have multiple models to cover different cases, we not go into this detail now, huge topic, sticking to our use cases.
JeffB: ok
Lenss: is the version of my diagram up to date?
LC: send me updated version, we can make sure.
Lenss: I think it is no web site,
LC: I’ll check intout.
LC: overview of credit. This is from MISMO, credit takes up a big chunk of MISMO. Highly automated between credit agencies and lenders exchanging date using MISMO.
Credit response means the credit report.
JeffB: difference between the information going in and the models doing the computation.
LC: yes. CRAs (Credit Reporting Agencies) act as middle men.
MaxG: probability of default
LC: yes, that is purpose of credit score.
JeffB: the number itself has no absolute meaning – only relative, just in a range of values.
LC: model to review:
LC: check definition of CreditReport
Max: report might not be reported by external agency, could be internal.
MU: what do we need to include about the origin of the credit report? The agency? the model? the computer program?
MaxG: models vary, a CRA might use multiple models. Each has strengths and weaknesses in different situations.
LC do we want to say what the entity was that produced the report, usually the CRA.
MU; lets keep it simple, 99% case is a CRA produced a CreditReport.
MU; look at CreditCheck
LC: you use INdependentParety for hasCreditReportingAgency, but AutonomousAgent produced the CreditReport. INdependentParty is subclass of AutonomousAgent.
MU: do we need to have CreditCheck, the actual request?
LC: maybe not, we focus on data, not the request.
MU: maybe I can merge the properties of the CreditCheck onto the credit report.
LC: not certain that we need this
MU: lets drop for now.
LC: lets go back to CreditReport. There are several types of products, e.g. fraud checking, residential mortgae report, reports for car loans.
JeffB: major CRAs, comes from broad mix. Data can come from a number of different sources. Maybe we want to catalog types of data sources, regardless of agency.
LC: yes. to MU: maybe want property sourcedFrom?
JeffB: sourcing from major CRAs, mostly have data from credit cards, that is original data, same data goes to all CRAs.
LC: yes, but what can you tell from data on the report? What matters is what is used for risk assessment. More confidence if more sources.
JeffB: source of report, not source of the data.
JeffB: could be product level.
MaxG: maybe credit is worthy of a separate stream? We could waste a lot of time doing things specific to loans. We should maybe factor out just what we need for loans. We are skating beyond loans scope a bit, must be careful.
LC: rating agencies a different beast.
MaxG: yes, very different customers, but they are giving a credit score of sorts.
LC: lets get back to HMDA. CRA is middle man, can
LC: forget about Risk Assessemt for the moment, lets get to credit report.
Max: credit report is a tool for risk assessment. This has attributes that we care about.
MU: that is what LC is trying to do.
Max: trick is where to stop.
LC: what do we need to describe the CreditReport? I suggest a property describing the provenance of initial data.
· borrower being assessed
·
MU: what kind of CreditReport are we talkina bout, comes from Experian etc?
LC: general. It will have product type, market domain, target audience, data source.
MU: 3 majors are the data repositories.
LC: roles: CreditRepositoryAgency (that produce reports).
LC: for mortgage, have middle man.
LC: CRA looks at all three data providers, and do an aggregation.
MU: ah, I see.
LC: I go away and draw this out in a figure.
LC: show the data used by MISMO.
MaxG: need information on quality of the reports, indicates how we use the report.
MU: this is a doing the work use case, HMDA is about reporting to regulators, not the doing of the credit assessment.
LC: next week I bring picture for Credit Reporting.
MU/LC TODO: look at HMDA spreadsheet for what they have on credit reporting.
Decisions
Action items
- Lynn Calahan [X] (Unlicensed) Michael
1. LC: create diagram of what is going on
2. LC/MU: look at HMDA spreadsheet for what they have on credit reporting
3. LC: look into what a credit use case might look like.