There has been a fairly emotional debate on this board about the scientific merit (or lack thereof) of credit scoring. For those of you who have an interest in similar methodologies here is a link describing http://www.tetrad.com/pub/prices/microvision.pdf MicroVision, a so called geo-demographic system used in direct marketing. In essence, it uses the data provided by the US census and feedback from merchants to categorize a particular ZIP+4 locale (normally 5 to 15 households) as one of the 50 market segments with names ranging from â??Upper Crustâ? to â??Movers and Shakersâ? to â??Bedrock Americaâ? to â??Struggling Metro Mixâ?. What I find fascinating about MicroVision is the sheer amount of marketing and behavioral information that can be extrapolated from census and other non-personalized data. Now imagine what sort of data mining can be done based on a set of very precise and by far more detailed information contained in CRA reports!
hi mkxt, Welcome and thank you! I'm interested and have only scrolled through it, going back for a real reading now. This is everyone's issue, I say. I was deeply touched by marci's postings today on the other thread. I for one, and I could be the only one, believe this should be a passionate issue for each of us. This is a brilliant board with some incredible minds, if we can all agree a thing is wrong, surely, we can come up with a way to make it right. "Disparate impact" that's such a clever phrase, call it what it is, discrimination. This link looks like it goes beyond the documentation from the other thread, it is putting everyone in a box with a zip code as a ribbon. Damn boxes. Sassy
LOL, That report refers to us (consumers) as zip+4's. ZIP+4's!!!!!!! I'm going to get myself a zip code that launches me into the "anomalies" box -- how useless their invasive report would be should we all do the same. And on any form requiring a disclosure of race or ethnicity, I'm checking that "other" box. Humphhhhhhhhhh, what a crock! Sassy (Zip+4 anomaly)
As individual as I like to think of myself, it was spooky how accurate their depictions are. I'm definitely a #13 (Successful Single) right down to the passport and everything. I guess this makes me a lemming? I'd love to see the database and find out what my zipcode is designated.
Sassy, thanks! Could you please post the link to that "other thread" you were referring to? I also think I should clarify a few things about my previous post: 1. I did not mean to imply that MicroVision is somehow part of the Fair Isaac scoring process (even though a similar approach may be used). I was just trying to illustrate how powerful data mining (with the good deal of imagination) could be. 2. I understand how sensitive this issue may be for a lot of people on this board, but I do not think that FICO scoring process is a bunch of bs. In fact, I think FICO is scientifically a very well founded econometric/statistical approach (even based on the little public information available about it). The problem I think lies in the fact that some people confuse scientific merits of scoring process with the adverse impact it may have on them. FICO scoring is designed for use by the CRA consumers - financial institutions! - not by the borrowers. And it certainly serves them (lenders) well. But who's to say that what's good for lenders is good for borrowers? 3. There were a lot of suggestions on the board for Fair Isaac to fully disclose its methodology, and beyond that, to provide answers to something like "So, how many credit cards, exactly, do I need to have to get a better score?" Without going to far in trying to explain why there might be NO right answer to that question, let me illustrate this. Suppose you are applying for a college admission and you are about to take an SAT. Now trying to get a better score, you call The College Board (organization responsible for SAT) and demand they give you the right answers upfront! However, if the College Board starts giving out right answers their services to consumers (colleges and universities, in this case) become worthless. If anyone can manufacture a perfect score what's the point of using the scoring process? In fact, SAT's went through a very similar controversy: a few decades back, college applicants were NOT allowed to see their test scores (just like FICO scores a few years back) reflecting the fact that the scoring was meant for colleges not for applicants. Now, we've come a long way since then, but up until today no one is giving away "the right answers" before you take the test. Hope this makes sense. mkxt
Hey everyone I am not quite sure where this falls under but I think if these people are doing it then others are as well. The single most dramatic product development in our recent history was the development of our "Preferred Zip Code" analysis software. It took four years to develop and test this unparalleled specialty tool which has helped propel Chase Credit to a leading position in this important market segment. Using this tool we have analyzed millions of three-repository merged files and took into account hundreds of attributes to develop various zip code preference tables based on a lender's underwriting criteria. The "Preferred Zip Code Tables" are perfectly adaptable for any industry requiring credit evaluation, such as automotive, credit card, tenant screening and even insurance. Our analysis could also guide a lender in determining the number of repositories to access based on the level of accuracy desired. Several national lenders are currently utilizing Chase's Preferred Zip Code Tables. the website where i got that is here: http://www.chasecredit.com/cra_general.htm
mkxt, No worries, I didn't associate it with FICO. At this first link below there is a link to a zipcode based boxing system as well. At the second link, is this from JohnM: "It seems that FICO does discriminate against a minority, those that are in a group that has a higher risk of default on a debt. If your credit habits fall out the â??normâ?, your score is dinged. FICO scores are good predictors of the credit worthiness of groups of people with similar habits. Any system of statistical prediction is useless for an individual case. That is the whole basis of statistical analysis. " JohnM I don't like the boxing and labelling of people, says this zip+4 anomaly! Now I'm going to go check out the link dario posted so I can find the "preferred zip code" and figure out how I can become a preferred anomaly ;-) I don't think FICO is b.s., nor am I confused about its scientific merits. Perhaps though we should start boxing and labelling the adverse impacts it does have on people based on the inaccurate information and assumptions being scientifically computed to spit out a number or a label for an individual based on those inaccuracies and assumptions. I understand as well who the CRA's work and exist for, besides their greedy pocket-lining organizational interests, but how can a system be serving the lenders well if it isn't based on accurate input? Garbage in = Garbage out. SAT's are referenced at the first link as well, I believe by clc. I don't have a problem necessarily with the idea of scoring, lenders ought to be making decisions based on something, but that something should have credibility as well. The scoring, labelling, and boxing of people and the resulting impact on individuals can't be sanitized by calling it a scientific meritous process. A computer is a box with guts that function based on data input. The resulting spit of a categorizing number or label that is then applied to individuals, whose economic lives are based on this magic spit, if not based on data with integrity, isn't worth the paper it was spit out on ;-) Fannie Mae is being sued: http://consumers.creditnet.com/stra...hreadid=33835&perpage=20&pgnum=1&pagenumber=1 The REASON for credit scoring: http://consumers.creditnet.com/stra...hreadid=15551&perpage=20&pgnum=1&pagenumber=1 Sassy
